“Check-Ride for Legal Writers”: Using Aviation Pedagogy to Rethink Assessment in the Age of Generative AI

Introduction

Law professors are divided as to whether generative AI^[1] represents a breakthrough or a threat to legal education. Some view it as a tool to enhance efficiency and expand access to legal resources,^[2] while others warn that its adoption risks eroding core skills and professional judgment.^[3] What is clear, however, is that this technology’s trajectory is fueled by enormous investment,^[4] underscoring AI’s inevitable permanence in society.^[5] As generative AI occupies a growing role in education, early research suggests it can destabilize students’ ability to read, write, and think critically, while creating an illusion of proficiency.^[6] Specific to legal writing courses and other skills-based instruction, this technology is disrupting core assumptions about skills development, authorship, and assessment.^[7] The challenge, therefore, is not only how to teach effectively with AI in the mix but also how to verify genuine doctrinal and substantive understanding and transferable skills.

Although this challenge feels novel, it is not without precedent. Generative AI’s integration into law practice parallels the aviation industry’s adoption of autopilot.^[8] Aviation pedagogy offers a cross-disciplinary perspective on when to introduce automation into a curriculum and how to evaluate human comprehension in automation-rich environments.^[9] A central feature of that pedagogy is the “check-ride,” an oral exam in which pilots must explain and defend their decisions before demonstrating them in practice.^[10]

Adapted for the law school classroom, a legal writing check-ride can serve the same function. In the fall semester, an oral exam can follow a draft memorandum to confirm analytical foundations before students finalize their written work. In the spring semester, an oral exam or an analogous exercise positioned between the written brief and the final appellate or trial argument can reinforce analytical skills. Together, these interventions respond to pedagogical and doctrinal challenges generative AI poses for legal education, including its implications for authorship, competence, assessment integrity, and professional responsibility norms.

A practical challenge, however, lies in scalability. Administering oral exams to large cohorts requires tools that extend faculty capacity. Here, the aviation industry’s development of autopilot provides a useful analogue.^[11] The very technologies that introduced new risks eventually generated the tools to manage them. Emerging AI tools capable of administering, recording, and transcribing oral exams suggest a similar dynamic in education. AI may create the need for new assessment models, but it may also supply the mechanisms that make them feasible. Aviation’s methods for cultivating judgment and confirming mastery amid increasing automation guide the legal writing check-ride model developed in this Article.

In doing so, this Article demonstrates how legal education can adapt longstanding assessment traditions to contemporary technological pressures without compromising rigor or fairness. This framework also aligns with the governing expectations for rigorous formative and summative assessment (ABA Standard 314)^[12] and supports professional identity formation (ABA Standard 303(b)(3)).^[13] Although developed in the context of first-year legal writing courses, the model translates well to doctrinal courses, clinics, simulations, and upper-level seminars.^[14]

Rapid advances in generative AI are reshaping every knowledge-based profession, including law, at a pace that leaves little room for complacency.^[15] In a recent dialogue with lawmakers at the National Governors Association 2025 Summer Meeting, entrepreneur Mark Cuban captured the urgency of this moment, observing that many aspects of modern education remain disconnected from the technological ecosystem that students now inhabit.^[16] Perhaps technology investors have created a “self-licking ice cream cone”—an emerging cycle in which the deployment of AI generates both new challenges and the tools to solve them.^[17] But as this reality unfolds, law faculty cannot afford to ignore these challenges or stand apart from the crucial conversations shaping AI’s role in legal education.^[18] The future of legal education will hinge on the academy’s ability to master teaching with AI, a task that demands creativity and adaptability.^[19]

This Article unfolds in three parts. Part I situates generative AI within the pedagogical traditions of legal writing and identifies the assessment challenges that arise when students can outsource core cognitive tasks. Part II draws on aviation pedagogy to show how another profession confronted the risks of automation, using the check-ride model to preserve human judgment amid increasing technological capability. Part III applies these insights to legal writing. It offers a detailed framework for implementing a check-ride-style oral exam, including its alignment with ABA Standards 314 and 303(b)(3), its role in professional identity formation, and practical considerations for design, rubrics, accommodations, scalability, and AI-assisted administration.

I. Generative AI’s Assessment Problem in Legal Education

To understand the stakes of this moment, this section examines the key pedagogical foundations of legal writing and the specific ways in which generative AI destabilizes long-standing assessment practices.

A. The Pedagogical Foundations of Legal Writing

Generative AI tools are now a routine part of law students’ academic lives, creating tension with a discipline whose foundations trace back to classical rhetoric.^[20] That tradition, however, has never been static. Bibliography courses in the 1920s focused on law library instruction and legal research skills training, but also included some, albeit limited, attention to legal writing.^[21] During the 1930s and early 1940s, these courses expanded to include broader training in legal skills, such as legal analysis through the case method.^[22] By the late 1940s, legal writing emerged as a distinct discipline, addressing the perceived need for basic writing instruction among post-World War II students.^[23] Over time, these courses evolved from teaching basic writing to prioritizing “finished legal writing.”^[24]

Finished documents, such as objective office memoranda or persuasive briefs, are the customary basis for summative assessment (i.e., the final, graded assignment) in today’s legal writing courses, unlike today’s doctrinal courses which typically culminate in a single final exam.^[25] Legal writing faculty use various assessment tools to teach students because they understand that effective coursework bridges the gap between the theory and practice of law.^[26] They use an integrated approach to learning by combining creative, real-world assignments and simulations, even as debates persist about what level of “practice-readiness” students should achieve.^[27]

Grading practices likewise receive sustained attention from legal writing faculty.^[28] Legal writing faculty take grading seriously because they recognize that grades provide essential feedback to students and serve as a universal marker of proficiency for prospective employers.^[29] Some legal writing faculty advocate for the use of rubrics to ensure consistency and transparency in grading.^[30] Others contend that rubrics stifle creative thinking and result in overly formulaic writing.^[31] The rise of generative AI has added a new layer to this conversation, raising questions about whether legal writing faculty can reliably evaluate student-authored work and assign meaningful grades.^[32]

B. Scholarly Perspectives on AI in Legal Education

These pedagogical foundations help situate the emerging scholarly conversation about how generative AI is disrupting legal writing instruction.^[33] John Bliss brings empirical rigor to the conversation, using national survey data to illuminate faculty concerns about AI in legal education.^[34] Faculty respondents expressed “grave concerns that students might ‘outsource’ their studies to AI and thereby miss important learning opportunities.”^[35] They reported uncertainty about how to approach assessment in this new environment.^[36] At the same time, faculty respondents understood that their concerns could not outweigh the reality that students must learn to navigate generative AI to be prepared for practice.^[37] Undoubtedly, in the ongoing effort to achieve efficiencies and reduce costs, law firms and in-house legal teams will adopt generative AI tools if they have not already.^[38] Students who are not literate in the ways of generative AI will be left behind.^[39]

Preparing practice-ready graduates, therefore, will require teaching students to use generative AI tools effectively and responsibly.^[40] Building on this point, Kirsten Davis examines more directly how legal writing pedagogy should evolve in the AI era.^[41] Her scholarship identifies the “big issues” confronting legal writing faculty at the dawn of this technological shift.^[42] She warns, “If generative AI can successfully write for students without their intellectual investment, this is a serious problem for legal writing courses. Legal writing courses rely almost exclusively on assessments produced outside of class in conditions where it would be tempting to use generative AI even if forbidden.”^[43] Davis concludes that “legal writing assessments will likely need to change in response to generative AI, and the field will need research and experimentation to this end.”^[44]

Lastly, Carolyn Williams offers important insights about the capabilities and limits of generative AI in the legal writing context.^[45] Her work elucidates what these tools can do, how prevalent they have become, and how unreliable detection mechanisms can be.^[46] She reminds readers that enthusiasm for generative AI should not obscure the reality that law students preparing for the NextGen bar exam must still demonstrate skills that cannot be outsourced to generative AI.^[47] For this reason, while she observes that “[m]any LRW professors already incorporate oral arguments into their 1L spring semester writing courses,” she encourages faculty to “consider adding alternative oral communication assessments, too.”^[48]

C. Three Challenges: Authorship, Overreliance, Skill Erosion

Surveying the existing scholarship reveals three main challenges for teaching legal writing effectively in an environment where AI tools are ubiquitous.^[49] First, generative AI complicates verifying student authorship.^[50] With a growing number of free and easily accessible tools, students may be tempted to use them, even when they violate course or university policies.^[51] Similarly, the boundaries between acceptable and impermissible use quickly blur in courses that permit limited AI use (for example, allowing students to use AI to edit and polish their writing, but not to generate substantive content).^[52] This concern goes to the heart of academic integrity and assessment validity.^[53]

Second, permitting AI assistance in research and drafting risks overreliance.^[54] While these technologies can meaningfully enhance legal research and writing, they remain imperfect.^[55] In her foundational contribution to the discourse on generative AI and law school, Williams highlights three critical concerns: “1) [Generative AI]’s tendency to hallucinate case law and legal sources; 2) its failure to recognize the truth; and 3) some analysis issues that apply to legal writing.”^[56] This overreliance can lead to writing that lacks accuracy.^[57]

Third, legal writing faculty worry that generative AI may discourage deep engagement with assignments.^[58] Unlike overreliance, which focuses on dependency for efficiency, this concern centers on the loss of authentic intellectual engagement and the formative process of skills development. Students who use generative AI as a “shortcut” or “crutch” risk failing to build core skills in reading, writing, and reasoning.^[59] After all, these are the very competencies legal writing courses are designed to cultivate.^[60]

D. Generative AI and the Illusion of Intelligence

To address these challenges, it helps to understand the nature of generative AI and its illusion of human-level intelligence. AI experts consider a machine to have achieved human-level intelligence if it can pass the “imitation game,” or the “Turing test.”^[61] British mathematician Alan Turing created the test in his renowned 1950 paper, “Computing Machinery and Intelligence.”^[62] Famously, he opened his paper with this question: “Can machines think?”^[63] But quickly thereafter, he pointed out some weighty philosophical challenges with this question.^[64] For the question to be meaningful, both the asker and the answerer must agree on what “thinking” means for a machine.^[65] Rather than wrestling with this abstract challenge, Turing took a more practical approach, considering that, in large part, human intelligence expresses itself through language.^[66]

Turing proposed that a machine can think if it can convincingly imitate human conversation.^[67] The traditional Turing test involves a human judge who interacts with two unseen participants simultaneously.^[68] Like the judge, one of the participants is a human, but the other participant is a machine.^[69] As the judge engages with each participant in a text-based conversation, the participants attempt to convince the judge that they are, in fact, human.^[70] If the judge cannot reliably identify the human participant, then the machine participant is considered to possess intelligence.^[71]

Many machine systems have been put to the Turing test in the past quarter-century.^[72] However, until the very new phenomenon of large language model (“LLM”)^[73] technology, none have passed.^[74] Recently, cognitive scientists determined that three LLMs (GPT-4.5, LLaMa-3.1-405B, and GPT-4o) could “pass the Turing test when they are given prompts to adopt a humanlike persona.”^[75] The study concluded that “it is arguably the ease with which LLMs can be prompted to adapt their behavior to different scenarios that makes them so flexible: and apparently so capable of passing as human.”^[76]

It is important to remember that the Turing test evaluates whether a machine can mimic human decision-making, not whether it understands the issues at hand.^[77] The judiciary, legal profession, and general public expect legal arguments and decisions to reflect human judgment.^[78] Thus, the ethical implications of machines capable of imitating human reasoning are far-reaching for the legal system.^[79] While LLMs may offer lawyers advantages in speed, consistency, and accuracy,^[80] the Turing test does not affirm that these technologies are capable of moral reasoning, which is a quality that remains essential to the integrity of the legal system.^[81]

II. What Aviation Already Solved About Automation and Assessment

Recognizing the challenges associated with generative AI in legal education, lessons from aviation pedagogy—another field transformed by automation—offer valuable guidance for legal writing faculty.^[82]

A. Autopilot and Human-Automation Dynamics

The challenges posed by generative AI are not unique to the legal profession. The aviation industry experienced its own technological reckoning with the introduction of autopilot systems that guide an aircraft’s trajectory without continuous human control.^[83] The analogy is instructive: generative AI is to legal writing as autopilot is to aviation.

No more than a decade after Orville and Wilbur Wright made their first sustained flight, Lawrence Sperry introduced the first aviation autopilot system in 1912; it used gyroscopes to control an aircraft’s pitch and heading.^[84] Today’s computerized autopilot systems did not become a feature of aircraft until the 1970s, and these systems now do much more than guide the aircraft’s in-flight path.^[85] Advanced autopilot systems can stabilize yaw (the aircraft’s rotation around its vertical axis), regulate flight speed, manage takeoff and landing, and assist with ground taxiing.^[86]

Some sources attribute the advent of autopilot systems to the 1950s and 1960s, when automation could manage basic stability functions such as roll, yaw, airspeed, and altitude.^[87] Modern autopilot, now called Automatic Flight Control Systems, did not emerge until the 1970s, when navigation and communication functions became integrated.^[88] Aviation automation matured slowly over decades, while LLMs have advanced at an extraordinary pace. Rapid progress, however, has not eliminated their limitations. In many respects, today’s LLMs resemble early cockpit automation because they are highly capable yet still require human monitoring and judgment.^[89]

Just as aviation could not ignore the advantages of autopilot despite the risks, legal education cannot avoid addressing AI’s expanding role. A 2011 article in Flying Magazine highlighted two pedagogically significant questions regarding autopilot systems: “Why use an autopilot? And, at what point should an autopilot be introduced into flight training?”^[90] If generative AI is transforming the role of legal writers, just as autopilot reshaped the role of pilots, both questions offer legal writing faculty clear parallels for inquiry: Why use generative AI? And at what point should it be introduced into a legal writing course?

To address the first question, it is helpful to examine the challenges and benefits of human-machine interaction. Three major challenges of automation in aviation are: (1) overreliance and complacency; (2) automation surprise; and (3) poor design by manufacturers.^[91] First, overreliance occurs when pilots depend too heavily on autopilot systems, a tendency the Federal Aviation Administration (“FAA”) warns “can lead to pilot complacency.”^[92] As the FAA notes, “The most insidious aspect of automation is its propensity to breed complacency and erode pilot confidence.”^[93] While the language is strong, the underlying logic is straightforward: prolonged reliance on autopilot can erode the manual skills required for safe flight.

Second, automation surprise occurs when a pilot does not fully understand the capabilities and limitations of an autopilot system.^[94] For example, the FAA notes that many autopilot systems “lack the ability to integrate aircraft position and terrain information,” meaning pilots may be surprised when the system does not maintain terrain or obstruction clearance.^[95] When an autopilot behaves unexpectedly or fails, the pilot’s workload increases dramatically, thereby heightening the risk of human error.^[96]

Third, poor autopilot design by manufacturers can create substantial risk, as illustrated by the worldwide grounding of the Boeing 737 MAX between 2019 and 2020.^[97] Boeing designed its Maneuvering Characteristics Augmentation System (“MCAS”) on the 737 MAX to compensate for aerodynamic changes caused by repositioned engines on the larger aircraft model.^[98] MCAS automatically adjusted the horizontal stabilizer to prevent stalls at high angles of attack.^[99] The system relied on only one sensor, however, and could repeatedly override the pilot’s commands.^[100] When faulty sensors sent erroneous data to MCAS during two flights, both aircraft entered uncontrollable nose-down dives.^[101] These failures highlight how design flaws in automation can have catastrophic consequences.^[102]

Even with these dangers, aviation’s history shows that automation offers substantial benefits. Autopilot systems reduce pilot fatigue and enhance safety by handling the routine yet attention-intensive aspects of flight.^[103] Without them, exclusive human piloting would demand uninterrupted concentration, increasing the likelihood of error.^[104] Autopilot also enables pilots to maintain situational awareness and devote more attention to higher-order decisions.^[105]

The timing of AI instruction, therefore, becomes a central pedagogical consideration. Aviation training again provides a helpful point of comparison. As general aviation instructors explain, “the best method is a graduated approach throughout the training curriculum” because “the first priority is to become comfortable with the hands-on control of the aircraft.”^[106] The same logic applies to legal writing. Students must first develop core competencies in research, analysis, and drafting without reliance on automation, so that they build the habits of attention and judgment that anchor professional practice.

Once students demonstrate reliable control of these foundational skills, instructors can begin to incorporate generative AI in a structured manner. Introducing AI first for formative, low-stakes assessments allows students to experiment with the technology while maintaining responsibility for the substance of their work. Its role can then expand incrementally, in parallel with students’ growing capacity for critical oversight. This stepwise progression mirrors the incremental training used in aviation and provides a coherent model for integrating generative AI into the legal writing curriculum. And just as aviation training requires pilots to demonstrate independent mastery even as cockpit automation becomes more capable, legal education must ensure that students can articulate and defend their analyses regardless of the technologies available to them. Understanding these parallels provides a foundation for exploring how aviation instructors teach complex skills in automation-rich environments.

B. Aviation Learning Theory

Aviation’s approach to training in the age of automation offers a useful blueprint for teaching complex skills in the age of generative AI. Historically, most law schools, dominated by the Socratic Method, gave little attention to cognitive learning theory.^[107] Yet, as most disciplines have recognized that cognitive learning theory is essential to course development and assessment, law schools have begun to follow suit.^[108] As discussed in more detail below, cognitive learning theory focuses on how learners acquire, store, and retrieve information.^[109] Legal skills training, including legal writing, draws from cognitive theory.^[110] Similarly, many techniques of flight instruction are grounded in cognitive theory.^[111]

Aviation pedagogy offers a structured framework for integrating complex skills through diverse learning methods and assessment strategies. Notably, the FAA, in conjunction with the U.S. Department of Transportation, publishes the Aviation Instructor’s Handbook, most recently updated in 2020.^[112] As described in its preface, the Aviation Instructor’s Handbook “provides aviation instructors with up-to-date information on learning and teaching, and how to relate this information to the task of teaching aeronautical knowledge and skills to learners.”^[113] It serves as a valuable lens for legal writing faculty, particularly as they seek to balance foundational skills development with the graduated incorporation of generative AI, while ensuring reliable evaluation of student outcomes.

The Aviation Instructor’s Handbook makes a considerable effort to describe the learning process. It explains that “all learning comes from perceptions, which are directed to the brain by one or more of the five senses: sight, hearing, touch, smell, and taste.”^[114] Factors affecting perception include “[p]hysical organism,” “[g]oals and values,” “[s]elf-concept,” “[t]ime and opportunity,” and the “[e]lement of threat.”^[115] Effective learners monitor how these influences affect their thought processes, allowing them to optimize mental strategies as they acquire new knowledge and skills.^[116] Similarly, effective instructors understand these dimensions and design instructional methods that intentionally support and enhance learning.^[117]

Building on this foundation, the Aviation Instructor’s Handbook also introduces Bloom’s hierarchy of learning objectives, originally developed by a team of educators led by Benjamin Bloom.^[118] This framework classifies six tiered levels of learning, each ranked by complexity: Knowledge, Comprehension, Application, Analysis, Synthesis, and Evaluation.^[119] On one end of the spectrum, Bloom considered Knowledge as the most basic form of learning because it merely requires the recall or recognition of facts or ideas.^[120] On the other hand, Evaluation involves assessing the value of information for a specific purpose.^[121]

In addition to featuring Bloom’s Taxonomy, the Aviation Instructor’s Handbook highlights other derivatives of cognitive learning theory, including Information Processing Theory and Constructivism.^[122] Information Processing Theory analogizes the brain to a computer to explain how humans gather, use, and store knowledge.^[123] While Bloom’s Taxonomy focuses on what students are doing cognitively, Information Processing Theory draws attention to how students think.^[124] A key aspect of this theory contemplates that, just like a computer, the brain has limitations.^[125]

Constructivism, as discussed in the Aviation Instructor’s Handbook, is based on the principle that knowledge is actively built through experience rather than passively received through instruction.^[126] It forms the foundation of higher-order thinking skills, which are central to both aviation instruction and legal education.^[127] Higher-order thinking skills involve the upper levels of Bloom’s Taxonomy: Analysis, Synthesis, and Evaluation.^[128] These levels go beyond memorization and recall, requiring students to think critically, exercise judgment, and make decisions.^[129] As the Aviation Instructor’s Handbook explains, “If the learner does not yet have much subject matter knowledge, they can draw on past experiences to gain entry into complex concepts.”^[130]

The Aviation Instructor’s Handbook illustrates this with the example of a child playing on a seesaw. Although the child has not formally studied concepts like weight or balance, the experience teaches them how the center of gravity works.^[131] Later, the child can draw on that foundational experience to understand more complex ideas, such as evaluating an aircraft’s weight and balance before takeoff. In this way, constructivist learning fosters meaningful connections between prior knowledge and new concepts.^[132]

To teach higher-order thinking skills effectively, instructors should employ delivery methods that go beyond traditional lectures.^[133] One such method, offered in the Aviation Instructor’s Handbook, is scenario-based training, which “uses a highly structured script of real-world experiences” to achieve learning objectives.^[134] According to the Aviation Instructor’s Handbook, scenario-based training setups “help learners better understand the decisions they have to make and also helps focus the learner on the decisions and consequences involved.”^[135] By placing students in evaluative, real-world contexts, scenario-based training promotes reflection, which lies at the heart of constructivist learning.^[136]

Taken together, the learning methods outlined in the Aviation Instructor’s Handbook underscore that lectures alone rarely create effective learning experiences. The most impactful teaching approaches incorporate inflection points that require decision-making and expose learners to the consequences of those decisions.^[137]

C. Assessment Models

To determine whether student pilots achieve proficiency in their craft, the Aviation Instructor’s Handbook prescribes two primary assessment methods: traditional and authentic assessment.^[138] Traditional assessment typically involves written tests, “such as multiple choice, matching, true/false, fill in the blank, etc. Learners typically complete written assessments within a specified time. There is a single, correct response for each item.”^[139] In contrast, authentic assessment “asks the learner to perform real-world tasks and demonstrate a meaningful application of skills and competencies.”^[140] As the Aviation Instructor’s Handbook explains:

Authentic assessment lies at the heart of training today’s aviation learner to use critical thinking skills. Rather than selecting from predetermined responses, learners must generate responses from skills and concepts they have learned. By using open-ended questions and established performance criteria, authentic assessment focuses on the learning process, enhances the development of real-world skills, encourages higher order thinking skills, and teaches learners to assess their own work and performance.^[141]

In practice, aviation training involves a structured progression before a student may even attempt the FAA’s formal assessments.^[142] Student pilots must complete required ground instruction, accumulate flight hours with a certified flight instructor, and demonstrate proficiency in key maneuvers, including solo flight.^[143] Only after satisfying these prerequisites may a student take the FAA written knowledge test and schedule the practical exam, commonly referred to as the “check-ride.”^[144] The written knowledge test exemplifies a traditional assessment that measures a student pilot’s recall and understanding of technical concepts under controlled conditions.^[145] The check-ride, by contrast, functions as an authentic assessment because it requires would-be pilots to apply their knowledge and skills in real-world scenarios.^[146] The check-ride consists of two parts: an oral exam and a subsequent flight test in an aircraft.^[147] By serving as a measure of competence and readiness, this multidimensional assessment process aligns with the goal of legal education to foster critical thinking and professional judgment. The most recent FAA Civil Airmen Statistics indicate that approximately 25 percent of private pilot applicants do not pass their initial practical test, and overall pass rates for initial FAA practical exams remain under 80 percent.^[148]

These interdisciplinary insights reveal that the assessment challenges posed by generative AI closely parallel those introduced by automation in aviation. Part III now applies these insights directly to legal writing courses, showing how a check-ride-style oral exam can test student learning while deepening higher-order reasoning skills.

III. Implementing the Check-Ride Model in Legal Writing

Generative AI will inevitably play a role in students’ writing processes, heightening the need for assessment methods that can verify individual understanding regardless of AI assistance.

A. Why Legal Writing Needs a Check-Ride Assessment

The automation-related dynamics described above re-emerge with particular force in legal writing. Just as aviation instructors confronted the pedagogical challenges introduced by autopilot, legal writing faculty now face comparable dynamics with generative AI. Like pilots, lawyers must excel at maintaining subject-matter expertise, exercising sound judgment, communicating clearly, remaining calm under pressure, adapting to changing conditions in real time, sustaining situational awareness, and attending to detail.^[149]

Used well, generative AI can handle repetitive or mechanical work, allowing legal writers to focus on substantive analysis. Scholars describe this as “Hybrid Human-AI” writing.^[150] In practice, AI can help organize research, summarize authorities, synthesize lines of cases, and proofread for consistency. At the same time, the lawyer retains responsibility for interpreting doctrine, drawing analogies and distinctions, and making normative choices.^[151] As Davis notes, “competent legal analysis and argument development might someday ethically require, not just permit, the use of a large language model in the prewriting stage,” because structured use can improve the scaffolding of analysis and the clarity of expression.^[152] Bliss further observes that responsible AI use “could lead to more effective lawyering and possibly enhanced well-being” in a profession long marked by stress and mental health strain.^[153]

Despite these benefits, automation threatens these competencies in predictable ways. Overreliance on generative AI can foster complacency, such as when legal writers allow a model to draft portions of a brief without independently verifying citations or reasoning.^[154] Automation surprise can arise if legal writers misunderstand the tool’s capabilities or limitations, such as assuming it can properly interpret jurisdiction-specific rules or ethical constraints. And design flaws in certain AI platforms can produce systemic errors, including fabricated case law or distorted statutory language.^[155] There are already too many reports of lawyers submitting documents to courts that contain hallucinated cases.^[156]

These automation dynamics have destabilized the assumptions about authorship and competence that once underpinned assessment in legal writing. Before generative AI, instructors could reasonably infer that a student who produced a strong brief and delivered a competent oral argument had both authored and understood the underlying analysis. That assumption no longer holds. Today’s tools can “answer study guide questions, write a term paper, produce a literature review, and do it more quickly than humans with almost no skill or effort required on the part of the student.”^[157] They can also “translate languages, compute mathematical calculations, and edit text for grammar.”^[158] As a result, legal writing faculty must remain aware “that some students, when given the opportunity, will rely on them to complete assignments in the least amount of time and with the least amount of effort.”^[159] Ensuring authentic learning now requires a multidimensional assessment strategy that tests not only the written product but also the student’s understanding of the analytical choices that produced it.

The legal writing check-ride method does exactly that. In a legal writing course, a check-ride assessment serves as a process-focused oral exam. It requires students to explain and defend the analytical choices underlying their written work. Rather than asking students to perform advocacy, the check-ride prompts them to articulate how they identified the governing rule, selected and synthesized authorities, structured their analysis, and applied law to fact. In this way, the check-ride verifies the student’s independent mastery of core competencies before any subsequent task, whether that task is finalizing the written product or preparing for simulated advocacy. The specific design features of the legal writing check-ride are discussed in greater detail in Part III(D).

B. Aligning Oral Exams with ABA Standards 314 and 303(b)(3)

The check-ride method aligns closely with the ABA’s standards concerning assessment and professional identity formation. ABA Standard 314 obligates law schools to use both formative and summative assessments to “measure and improve student learning and provide meaningful feedback to students.”^[160] Under the ABA’s interpretation, formative assessments offer feedback at various points to improve learning, while summative assessments evaluate the degree of learning at a course’s culmination.^[161]

Strikingly, the FAA articulates these same concepts with clarity and pedagogical purpose. The Aviation Instructor’s Handbook explains that formative assessments “provide a wrap-up of the lesson and set the stage for the next lesson.”^[162] In contrast, summative assessments “measure how well learning has progressed to that point.”^[163] In both frameworks, formative assessments guide improvement, and summative assessments verify proficiency.

So, what would a check-ride in the first-year legal writing course look like? In the fall semester, an oral exam, positioned after a draft but before the final submission, serves as a potent formative assessment. It allows legal writing faculty to identify analytical gaps, clarify misunderstandings, and provide “meaningful feedback to improve student learning.”^[164] By requiring students to articulate the reasoning behind their written choices, the oral exam reinforces foundational skills in research, analysis, and organization at a critical midpoint in the writing process.

In the spring semester, an oral exam can serve as a summative assessment by requiring students to defend their analytical choices after submitting their finished writing product. This form of evaluation directly “measure[s] the degree of student learning” by testing whether students can articulate, justify, and apply the legal principles reflected in their brief.^[165] It also provides a reliable mechanism for determining whether the final written work represents the student’s own understanding. In aviation, competency reflects far more than a minimal threshold. FAA practical test standards require candidates to demonstrate judgment, adaptability, and clear articulation of the reasoning underlying their decisions. A legal writing oral exam functions similarly by assessing not only the accuracy of a student’s analysis but also the quality of the reasoning that supports it.

The check-ride model also aligns with the ABA’s professional identity formation requirement. ABA Standard 303(b)(3) requires law schools to create substantial opportunities for students to reflect on the ideals of the profession.^[166] As the ABA notes, this process demands “intentional exploration” of the principles and best practices that are foundational to competent and ethical lawyering.^[167]

Students today are navigating an evolving professional landscape in which powerful AI tools can unsettle them. Generative AI enables students to outsource analytical effort, thereby weakening the reflective habits that underlie professional identity formation.^[168] Additionally, although some students may appreciate the efficiency and promise of generative AI, they may worry that these tools could diminish the value of their own emerging professional abilities.

Scholars studying professional identity development generally agree that it rests on three interrelated cognitive and behavioral capacities: (1) metacognition, (2) self-regulation, and (3) self-efficacy.^[169] Metacognition refers to a person’s deliberate awareness of how they think.^[170] Self-regulation involves the strategies individuals use to initiate, maintain, and adjust their thoughts, actions, and emotions as they work toward defined goals.^[171] Self-efficacy describes an individual’s belief in their own ability to plan and take the steps necessary to achieve those goals.^[172]

The check-ride method helps students strengthen each of these capacities. First, it promotes metacognition by requiring students to articulate why they framed an issue a certain way or how they interpreted an authority. Second, the conversational format compels students to take ownership over their learning, manage uncertainty during the dialogue, and regulate their emotional responses to challenge. Finally, the check-ride setting gives students a structured opportunity to demonstrate their competence in real time, helping them become self-assured in their ability to perform the intellectual tasks lawyers routinely face.

Periods of technological upheaval have historically served as inflection points for professional identity formation.^[173] In this new environment of AI tools, law students may struggle to discern the limits and the strengths of their own knowledge and instincts, making it harder to develop confidence in their independent judgment.^[174] As commercial aircraft became increasingly automated, pilots found that while automation reduced routine workload, this made confidence in higher-order judgment, situational awareness, and the ability to manage the unexpected more essential than ever.^[175] The check-ride model offers a parallel in legal education by anchoring professional identity in demonstrated independent reasoning under pressure.^[176] Accordingly, beyond its evaluative role, the check-ride experience strengthens the self-confidence and autonomous judgment necessary for a mature professional identity.^[177]

C. Historical and Pedagogical Foundations of Oral Exams

To understand how check-ride-style oral exams can strengthen legal writing instruction in the age of generative AI, legal writing faculty should examine the historical foundations, pedagogical benefits, and practical considerations associated with this form of assessment. Oral exams were once the gold standard of academic assessment.^[178] But as public education expanded beyond the mid-1800s, the logistical demands of mass schooling led educators to favor more efficient methods.^[179] Early standardized tests quickly replaced oral assessments as the dominant tool for evaluating student performance.^[180] While this shift brought undeniable benefits, including broader access to education across diverse and growing urban populations, it also brought drawbacks.^[181]

The mass administration of written exams led one expert to analogize it to the production of goods, where “examinations were the means of judging the value added to the raw material . . . during the course of the year.”^[182] This factory-line approach incentivized instructors to align course materials too closely with test content and over-emphasize rote memorization as a skill.^[183] In turn, public education began to diverge from true skills-based models rooted in apprenticeship and experiential learning, which relied heavily on dialogue, demonstration, and individualized feedback, much like oral exams.^[184]

The salient point is that many institutions abandoned oral exams for logistical rather than pedagogical reasons.^[185] Nevertheless, it would be imprecise to treat oral exams as unearthed fossils no longer alive in contemporary American education. To the contrary, some disciplines continue to use them wherever logistics still allow.^[186]

Additionally, oral exams continue to play a role in European legal education.^[187] For example, John Burman described his experience teaching at Petrozavodsk State University in Russia, where “[o]ral exams are the traditional method of testing students” and usually the only method of assessment.^[188] He explained:

Five students enter the classroom while the others wait in the hall. Each chooses one question from about fifty . . . . Each student is then allowed time to think about the answer (without books or notes). As soon as one of the five is ready, she meets with the teacher. After she answers the written question, along with any followup questions, the teacher assigns her a grade, which is then entered into the student’s grade book . . . . Those students who do not pass muster the first time around may return at the end of the session and try again.^[189]

Burman observed several advantages to this system. It fosters students’ ability to articulate their knowledge and think on their feet, provides opportunities for real-time self-assessment, and makes it difficult to mislead the examiner—“[I]t’s hard to snow someone for more than a couple of minutes,” he remarked.^[190] He further observed that when oral exams serve as the sole method of assessment, they relieve professors of the burden of grading written work.^[191] These observations, as the next section will show, are deeply aligned with the educational objectives that oral exams seek to advance in legal writing courses.

Burman advocates for a combination of oral exams and written assessments as the most effective approach to evaluating student performance.^[192] When he returned to teach at the University of Wyoming, he redesigned his upper-level required Professional Responsibility course to include both written and oral components in the final exam.^[193] Reflecting on this experience, he stated that “a system of evaluation that relies primarily on one system or the other benefits some and penalizes others.”^[194] His perspective supports the premise that oral exams can bridge traditional written assessments and simulated oral arguments in legal writing courses, and that using these methods together yields a more comprehensive measure of student learning.

Both then and now, across disciplines and around the globe, oral exams have been a favored method of assessing student performance for many reasons.^[195] Before the assessment, oral exams promote active preparation and collaboration. Students must anticipate questions and develop responses, often working with peers to exchange ideas and practice delivery. This process strengthens their interpersonal and communication skills and increases motivation to engage with the course material.^[196] Students report that oral exams reinforce their expertise in the course material and that they are more motivated to engage with the course material over the specter of an oral exam than that of a written exam.^[197]

During the assessment, oral exams provide immediate, interactive feedback.^[198] The format requires students to communicate directly with the examiner, allowing for clarification, guidance, and elaboration. Rather than being penalized for a misunderstanding or incomplete answer, students are encouraged to refine their reasoning in real time.^[199] After the assessment, students are more likely to internalize and apply feedback because oral exams demand presence and engagement.^[200] Oral exams also help students develop a sense of ownership over their learning, which extends beyond the exam itself and contributes to the professional identity they will carry into legal practice.^[201] This is especially important in an era of generative AI, where students may be tempted to bypass the intellectual effort required to internalize and explain ideas. That shortcut can weaken their connection to the material and, in turn, their self-efficacy.^[202]

Oral exams, in effect, guarantee that students confront and work through the intellectual challenges essential to forming a confident and coherent professional identity. They offer a pedagogically rich supplement to written assessments, supporting learning at every stage and resisting the passive tendencies that AI-generated work can encourage. Oral exams, however, are not without drawbacks. This Article acknowledges those challenges while identifying practical measures for mitigating them and preserving the distinct pedagogical value oral exams offer.

D. Designing Oral Exams for Legal Writing Courses

The design and administration of exams require careful attention to both structure and evaluation criteria.

1. Structure and Sequencing

The first step in redesigning the summative assessment process for a legal writing course is determining where to place the oral exam. Most programs emphasize objective writing in the fall and persuasive writing in the spring, with the fall semester culminating in an objective office memorandum and the spring concluding with a persuasive brief and a simulated oral argument.^[203] Although each school structures its curriculum differently, this progression is broadly consistent across institutions.^[204]

To strengthen assessment in the spring, legal writing faculty should integrate an oral exam that is distinct from the traditional oral argument. Although students may participate in practice oral arguments as preparation for the simulated argument, these exercises serve a different pedagogical purpose. A practice oral argument rehearses advocacy and courtroom presentation, while the oral exam evaluates the student’s underlying analytical reasoning and decision-making independent of advocacy performance.^[205] Returning to aviation pedagogy as a helpful analogy, flight training sequences a written knowledge test before the check-ride, which begins with an oral exam and ends with a practical flight test.^[206] Applying this model to legal writing courses, the written brief serves as the counterpart to the written exam, while the simulated oral argument parallels the flight test. The oral exam, positioned between them, serves as the bridge connecting students’ written analysis to their real-time reasoning and judgment.

A similar structure works in the fall semester. Because no oral argument component typically exists in the fall, the oral exam can occur after students submit a good-faith draft of their memorandum.^[207] This timing allows professors to probe research and drafting choices and to provide meaningful formative feedback before the final submission.

2. Scenario-Based Questioning

With the sequencing established, the next consideration is the structure of the oral exam. The overarching learning objectives of a legal writing course include developing both critical thinking and practical lawyering skills. In other words, legal writing instruction aims to cultivate higher-order thinking skills, which the Aviation Instructor’s Handbook defines as “cognitive processes such as problem solving and decision-making, as well as the cognitive skills of analysis, synthesis and evaluation.”^[208] Further, the Aviation Instructor’s Handbook outlines several criteria for crafting effective oral exam questions that promote higher-order thinking.^[209] Questions should relate directly to the subject matter, be clear and concise, match the learner’s ability and stage of training, focus on a single concept rather than multiple ideas, and present a meaningful challenge to the learner.^[210] The Aviation Instructor’s Handbook advises against ineffective question types in oral exams, such as yes/no questions, overly complex or confusing scenarios, trick questions, and those unrelated to the subject matter.^[211] These types of questions can undermine learning by distracting students, creating unnecessary confusion, or shifting focus away from the instructional goals.^[212]

When developing questions, a good starting point is the course learning objectives. At Duquesne Kline School of Law, where I teach legal writing, the fall semester first-year course centers on four high-level objectives: (1) Legal Research, (2) Legal Analysis, (3) Legal Writing, and (4) Other Skills Necessary for the Ethical Practice of Law.^[213] Specific learning outcomes support each objective.^[214] For example, under “Other Skills Necessary for the Ethical Practice of Law,” one outcome includes submitting work on time and in accordance with formatting rules.^[215] Here are some examples of higher-order thinking questions tailored to each objective:^[216]

Legal Research

• How can a researcher determine whether a source is a controlling authority for the jurisdiction, and why is that distinction critical for objective analysis?

• What steps should be taken to verify the accuracy of a statutory citation before including it in a memorandum?

• In what ways can secondary sources shape the direction of primary source research, and what are the risks of overreliance on them?

• What strategies ensure that research captures both favorable and unfavorable authority, and why is this balance essential for objectivity?

• How should a researcher confirm that case law is still good law, and what tools support this process?

Legal Analysis

• What factors should guide the inclusion of dicta in an analysis, and how can its significance be accurately represented?

• How can a writer explain a plurality opinion in a way that reflects its limited precedential value without overstating its authority?

• What steps ensure that policy considerations are presented neutrally rather than as advocacy?

• How should conflicting lines of authority be synthesized to provide a balanced and accurate prediction of how a court might rule?

• How should a legal writer evaluate the weight of persuasive authority from another jurisdiction, and what factors determine whether it should be included in an objective analysis?

Legal Writing

• How does the analytical structure promote clarity and neutrality in an office memorandum, and when might adjustments be necessary?

• What criteria should guide whether a case is presented as a full illustration, an explanatory parenthetical, or a simple citation in objective writing?

• How can sentence structure and word choice be revised to eliminate ambiguity and maintain an impartial tone?

• What characteristics make a topic sentence effective in signaling the purpose and scope of a paragraph, and how can it guide the reader through complex legal analysis?

• How can transitions between rule explanation and application be crafted to maintain logical flow and objectivity?

Other Skills Necessary for the Ethical Practice of Law

• What professional steps should a writer take when discovering an error in submitted work, and how should the correction be communicated?

• How can confidentiality be preserved when using AI tools for drafting, and what ethical rules govern this process?

• What strategies help prevent inadvertent plagiarism when paraphrasing case law or statutes?

• How should a writer manage time to meet deadlines without sacrificing accuracy and completeness?

• What are the ethical implications of omitting unfavorable authority in an office memorandum, and how should this be addressed?

It is up to each legal writing professor to develop questions that align with their own course objectives and learning outcomes, but the examples provided here illustrate possible approaches. The number and scope of questions may vary depending on factors such as exam length, course emphasis, and instructional goals.^[217]

Beyond these types of questions, the Aviation Instructor’s Handbook identifies scenario-based training as a primary method for teaching advanced cognitive skills, emphasizing its effectiveness in bridging theoretical knowledge and practical application.^[218] Incorporating scenario-based training into oral exams aligns with established methods for cultivating higher-order thinking because it requires learners to apply concepts in realistic contexts rather than recite rules in the abstract.^[219]

Legal writing faculty can adapt this approach by designing scenarios that correspond to specific course learning objectives. For example, this scenario targets the Legal Research and Legal Analysis objectives:

A first-year associate at a law firm is asked to draft an office memorandum analyzing whether a client has a viable claim for false imprisonment. The associate has never handled a false imprisonment case and is not familiar with the elements of the claim. What steps should the associate take to identify the governing law and formulate an accurate Question Presented?

This exercise reinforces the Legal Research objective (locating authoritative sources) and the Legal Analysis objective (applying those elements to the client’s facts). This scenario has no single correct answer, nor does it suggest an obvious one. A reasonable response might identify the associate’s first step as consulting reliable secondary sources, such as a treatise or restatement, to identify the elements of false imprisonment. Another valid place to start might involve interviewing the client to gather relevant facts. Even with little to no facts, a student can still explain the purpose, format, and mechanics of a Question Presented.

The scenario can then be scaffolded to build complexity and address additional objectives. For example, this question builds on the prior one but focuses on the Legal Writing and Other Skills Necessary for the Ethical Practice of Law objectives:

After researching the memorandum, the associate is ready to begin writing. The associate located two cases that reach different conclusions on a key issue. One is from an intermediate appellate court, and the other is from the state’s highest court. Which case should the associate cite in the memo, and what steps should the associate take to ensure the research is accurate and complete before finalizing the analysis?

By layering questions in this way, professors can move students from foundational skills to advanced application. The open-ended nature of this oral exam exercise and its capacity to scaffold promote hallmarks of higher-order thinking skills, such as analytical reasoning, strategic decision-making, and metacognitive reflection.^[220]

Legal writing faculty need not limit scenarios to third-person framing; they can also use first-person prompts (e.g., “You are an associate at a law firm. How would you . . . ?”) to create authentic assessments that mirror real-world decision-making. Similarly, oral exams can be framed creatively; for example, an exam might take the form of an “update meeting,” in which the student assumes the role of an associate providing an update to the legal writing faculty acting as the managing partner. Alternatively, an oral exam can take the form of a “client consultation,” in which the student explains a legal concept or process to a “client” (played by the professor).

3. Rubrics for Fair and Effective Assessment

Equally important is the development of fair and transparent evaluation methods that measure substantive knowledge, analytical reasoning, communication skills, and professional judgment. Broadly, a rubric is “a framework for successful student learning by merging the criteria for the goal with a rating scale.”^[221] There are two types of rubrics. Holistic rubrics evaluate a work product using a single overall judgment, while analytical rubrics assess multiple outcomes or dimensions of the work separately.^[222] Calculation methods can be either quantitative, relying on numerical scores, or qualitative, relying on descriptive ratings.^[223]

Beyond the type of rubric and calculation method, professors must determine performance criteria. To do so, they should: (1) identify the intended learning outcomes; (2) determine the work product to be assessed; (3) specify the skills students must demonstrate in that work product to meet the learning outcomes; and (4) articulate performance indicators that reflect varying levels of achievement on those outcomes.^[224] In the end, performance criteria should be student-centered, measurable, and clear.^[225]

In the course of academic inquiry, three scholars stand out for their insightful contextualization of the use of rubrics in legal writing instruction. Jessica Clark and Christy DeSanctis endorse rubrics as a valuable tool in legal writing instruction.^[226] They assert that using rubrics sets expectations for students and promotes consistency in assessment among instructors teaching the same course.^[227] In this way, rubrics also serve as training material for the assessor.^[228] Although Clark and DeSanctis acknowledge that rubrics alone may not provide sufficient feedback, they note that rubrics are, by and large, an effective tool for identifying a student’s strengths and weaknesses.^[229]

Debrah Borman takes a harsher stance on the use of rubrics as an assessment tool.^[230] Her central critique is that rubrics are too rigid, describing them as “inauthentic pigeonholing that ‘stamps standardization’ onto a creative and analytical . . . process.”^[231] To Borman, rubrics impose a one-size-fits-all framework on complex subject matter and metacognitive learning processes.^[232] Additionally, she contends that rubric-based grading can undermine critical thinking by shifting students’ focus away from authentic learning and toward a check-the-box mindset.^[233]

Interestingly relevant to the topic of oral exams, Borman advocates for oral feedback as an alternative to rubrics.^[234] According to this scholar, “oral feedback provides the best environment for observations, questions, clarifications, and responses.”^[235] She further argues that “[o]ur holistic feedback should include oral collaboration . . . to encourage important development.”^[236] Her emphasis on holistic evaluation aligns with the views of many experienced legal writing faculty.^[237] As the Legal Writing Sourcebook explains, effective assessment requires “flexibility to consider unexpectedly good (or bad) thinking and writing and reward (or penalize) it accordingly, without the constraint of a pre-existing grading rubric/checklist.”^[238]

Upon reviewing this scholarship, several questions emerge regarding how to assess oral exam performance. As a threshold matter, legal writing faculty must decide whether to evaluate the oral exam with a rubric or to treat it as a grade-by-completion exercise. Legal writing faculty should use a rubric. Unlike a completion-based approach, a rubric enables meaningful quantitative or qualitative assessment in a high-stakes setting and provides a structured, transparent measure of student abilities.

Faculty must then determine its structure, including the rating scale to be applied and the performance criteria to be assessed. At one end of the spectrum is a pass/fail system as the simplest form.^[239] At the other end lies a fully numeric approach, in which oral argument counts as a discrete percentage of the course grade and receives a quantitative score.^[240] Between these two systems are alternative models.^[241] For example, professors might adopt a competency-based scale with categories such as “highly competent,” “competent,” and “developing.”^[242] Legal writing faculty should select a scale that aligns logically with the course’s overall design and assessment philosophy.

Legal writing faculty seeking guidance on this topic may benefit from interdisciplinary insight using the Aviation Instructor’s Handbook.^[243] It outlines five gradations in its rubric for assessing flight training maneuvers: Not Observed, Describe, Explain, Practice, and Perform.^[244] At the Describe level, “the learner is able to describe the physical characteristics and cognitive elements” of a concept.^[245] In a legal writing context, this might resemble a student explaining the structural features of a predictive legal memorandum. At the Explain level, “the learner is able to describe the scenario activity and understand the underlying concepts, principles, and procedures that comprise the activity.”^[246] However, they may still require assistance to perform it.^[247] This could be a student who understands basic research techniques but struggles to select and apply relevant cases. At the Perform level, “the learner is able to perform the activity without instructor assistance.”^[248] This is a student who can independently and competently draft a legal memorandum. The Not Observed category is reserved for “any event not accomplished or required.”^[249]

The Aviation Instructor’s Handbook employs an analytical rubric to assess various aspects of flight training maneuvers.^[250] In the leftmost column, the rubric lists specific maneuvers (e.g., steep turns, slow flight, stalls, emergencies).^[251] The next column is used to record the performance gradation (Describe, Explain, Practice, Perform, or Not Observed).^[252] A third column provides space for comments.^[253] Most notably, the Aviation Instructor’s Handbook recommends that both the learner and the instructor complete the rubric.^[254]

This shared assessment system offers two key advantages. First, it actively engages students in the assessment process, fostering habits of reflection and self-evaluation.^[255] Second, the Aviation Instructor’s Handbook emphasizes that performance gradations are not linked to self-esteem or prestige, but rather to demonstrated levels of competence.^[256] This framework offers valuable guidance to legal writing faculty developing oral exam rubrics that emphasize skills development over subject-matter expertise. While subject matter expertise remains important, legal writing courses are fundamentally designed to cultivate critical thinking and core legal skills.^[257]

Because the oral exam is intended to assess students’ critical thinking and legal reasoning skills, it should be designed to evaluate whether students have met predetermined learning outcomes independently of AI assistance. The items tested should reflect the course’s stated objectives and learning outcomes, ensuring alignment between assessment and instructional goals. Furthermore, legal writing faculty must determine appropriate procedures for students who do not initially demonstrate the required level of competency. In such cases, students should be allowed to retake the exam after targeted remediation.^[258] The aim is developmental, not punitive.

4. Mitigating Anxiety, Bias, Accessibility, and Logistics

While oral exams offer many benefits, legal writing faculty should be aware of their limitations.^[259] These concerns include subjectivity in grading, student anxiety, accessibility challenges, and logistical difficulties.^[260] But these challenges can be mitigated and should not deter legal writing faculty from incorporating oral exams into their assessment processes.^[261]

Oral exams can be highly subjective,^[262] and this discussion does not ignore that reality. Subjectivity allows examiners some flexibility in assessment, but it also carries the risk of inconsistent grading, as different examiners may apply varying standards.^[263] More than that, examiners may be influenced, either consciously or unconsciously, by a student’s appearance or prior performance, neither of which reliably reflects the student’s current ability.^[264] This involves the halo effect, which examiners should avoid.^[265] The halo effect occurs when an examiner forms a positive or negative impression of a student based on observed traits or performance, and then extends that judgment to other aspects of the student’s ability, whether observed or not.^[266] The potential for bias in oral exams can be reduced by using multiple assessors, thereby balancing individual perspectives.^[267] Furthermore, as discussed in the previous section, carefully constructed rubrics promote fair and consistent evaluation across examiners.^[268]

Anxiety induced by the oral exam format may cause some students to underperform, even when they have a strong grasp of the material.^[269] To help students manage performance anxiety during oral exams, legal writing faculty can implement the following strategies:

• Clearly outline oral exam requirements and grading criteria at the start of the course.^[270]

• Provide practice oral exam sessions to help students build confidence.^[271]

• Share a live or recorded mock oral exam so students can become familiar with the format.^[272]

• Remind students that some anxiety in high-stakes situations is normal and, in moderation, can enhance performance.^[273]

When oral exams are thoughtfully incorporated into coursework, they not only provide an authentic assessment but also help students develop strategies for managing performance anxiety.^[274]

Beyond anxiety, legal writing faculty should recognize that certain students may face additional challenges when taking oral exams.^[275] For example, students who speak English as a second language may find it more difficult to process and respond to questions on the spot.^[276] Additionally, students with speech or auditory processing disabilities may struggle with real-time communication.^[277] Yet, oral exams may be preferred by a subset of students because they are “more inclusive and better serve some students with certain disabilities, such as dyslexia.”^[278] To ensure fair evaluation, professors should work with their institutions to provide reasonable accommodations.^[279]

Suzanne Rowe has conducted remarkable research on accommodations in legal writing courses, and some of her findings are notable here in relation to oral exams. She observes that “[o]ral communication is an essential lawyering skill, and many legal writing classes include some form of formal oral exercise.”^[280] Thus, it is not unreasonable to require students to complete an oral exam.^[281] She urges schools not to waive such requirements:

If students with learning disabilities try to avoid oral presentations, they should be reminded that even attorneys who do not pursue careers in litigation or appellate advocacy may find themselves making presentations to boards of directors or legislative committees. Thus, a school generally should not waive the requirement . . . , but some accommodations may be appropriate.^[282]

Rowe offers several accommodations options:

• Students with learning disabilities may receive extra time during the exam session to process and respond to questions.^[283]

• If the examiner is someone other than the professor, the examiner should be apprised of the student’s disability so they can exercise patience.^[284]

• Oral exams should be administered individually, without an audience, for students who stutter or experience panic attacks.^[285]

• If time and resources allow, professors may permit students with disabilities to complete additional practice oral exams to familiarize themselves with the setting and expectations.^[286]

Rowe emphasizes that “[t]hese accommodations need to be accompanied by a reminder to the student that real judges may not be patient and extra time may not be allowed.”^[287]

Finally, and perhaps most challenging, are the logistical complexities of administering oral exams to large groups of students.^[288] While time-consuming and difficult to manage, several strategies can help:

• Conducting oral exams in small groups can reduce scheduling burdens and allow for more focused interaction.^[289]

• Using tools like Zoom or Microsoft Teams offers flexibility in scheduling and can accommodate remote participation.^[290]

• Maintaining a structured set of questions helps guide the dialogue and ensures efficient use of time.^[291]

• AI can be leveraged to support the efficient and scalable administration of oral exams.^[292]

The obstacles that oral exams present are not insurmountable, and their benefits often outweigh their drawbacks.^[293] With thoughtful planning, legal writing faculty can harness the benefits of oral exams to promote authentic assessment.^[294]

E. Distinguishing Oral Exams from Oral Arguments

Some readers may wonder how an oral exam differs from an oral argument. Although both involve spoken communication, oral exams and oral arguments serve fundamentally different purposes.^[295] An oral argument is the capstone experience in most legal writing courses and is widely regarded as a best practice in legal writing pedagogy.^[296] It evaluates a student’s oral advocacy skills and their grasp of the facts and law related to the assigned problem.^[297] Through this exercise, students develop courtroom presence, persuasive speaking abilities, and the capacity to “think on their feet” by responding to questions from mock judges.^[298] In essence, oral arguments are centered on the substance of the case and teach students the “nuts and bolts” of courtroom advocacy.^[299] The goal of an oral exam, however, is to capture and differentiate students’ “understanding of the larger concepts included in the course.”^[300] Legal writing faculty should therefore incorporate oral exams, not as a replacement of oral arguments, but as a complementary method of assessment.^[301]

F. AI-Assisted Models for Scalable Exams

Recent data shows 196 law schools with an average first-year enrollment of about 223 students in 2025.^[302] For one law school with 223 first-year students, administering 10-minute oral exams to each student would require more than four full workdays. Extending the exam to twenty minutes per student would take over nine workdays, and a 30-minute exam would take nearly fourteen workdays.^[303] Data on the average number of legal writing faculty per school is sparse, but assuming most programs fall within the range of three to six full-time faculty members, a fifteen-minute oral exam for 223 students would require about seven faculty workdays in total—roughly two to three workdays per professor if three share the load, or just over one workday each if six professors participate. Of course, these estimates do not account for breaks, lunch, or time devoted to other classes and responsibilities, which would significantly increase the actual time required.

Given these statistics, the logistical limitations of oral exams become clear. In response, three AI-supported models offer potential solutions for these constraints.

Model 1: Transcript-Based AI Oral Exam. In this model, an AI avatar serves as the examiner, following a workflow similar to a written exam. Students would upload their written document before the exam, which the program would use to generate tailored questions. Students would then record oral responses to the AI’s prompts. The system would transcribe and organize these responses so the professor could grade them as if they were written answers.

Model 2: Real-Time AI Oral Exam. Similarly, this model uses an AI avatar examiner to administer the oral exam. After uploading their written document, students would orally respond to tailored questions, and the AI program would apply a rubric to generate a provisional score. The professor would then review flagged responses or samples to ensure quality.

Model 3: Adaptive AI Oral Exam. In his talk to the National Governors Association, Mark Cuban suggested that future assessments might combine face-to-face interaction with an “AI-based module where you are continually responding.”^[304] He considered that students might try to “go out to a third-party large language model to try to come up with the answer,” but the module would counter this by continually “raising the bar” to challenge students.^[305] In essence, students would be evaluated based on how far they advance through progressively more difficult levels of the assessment.^[306]

Treating these models as a kind of thought experiment allows us to probe the pedagogical and logistical boundaries of what AI-supported oral exams might achieve. One near-fault-proof way to prevent student cheating is to record students during the exam to verify identity and ensure they are not receiving outside assistance. The system could also capture screen activity to confirm that students are not using third-party generative AI tools.

Although other models surely exist, these three examples collectively demonstrate the feasibility of AI-proctored oral exams. Upon reviewing these models, it becomes clear that AI-administered oral exams offer notable advantages over traditional in-person oral exams:

• Analytical rigor. AI-assisted oral exams can be programmed to prioritize analytical depth over presentation style.

• Anonymity. AI-assisted oral exams keep students anonymous to professors, minimizing halo effects.

• Scalability. AI-assisted oral exams can be administered asynchronously or remotely, reducing scheduling burdens.

Beyond these shared benefits, legal writing faculty should weigh several key considerations when selecting a model:

• Budgetary resources. Dynamic or adaptive systems, such as Model 3, may involve higher development and maintenance costs.

• Institutional policy constraints. Legal writing faculty must determine to what extent school policy allows for AI-assisted grading.

• Level and timing of feedback. Model 2 and Model 3 provide immediate or near-immediate feedback, supporting formative learning; Model 1 involves delayed feedback because responses must be transcribed and graded like a written exam.

• Objectivity vs. human oversight. Automated rubrics enhance consistency and efficiency but may reduce nuanced judgment; automation can also help minimize unconscious bias.

• Time constraints. AI oral exams are faster than in-person methods, but Model 1 may require more grading time than Models 2 or 3 due to its transcript-based workflow.

• User experience. Legal writing programs must also consider technical infrastructure and student experience to ensure accessibility and fairness.

As a final matter, the reader might reasonably ask whether an adaptive written assessment could achieve the same benefits as an oral exam. At first glance, a traditional written exam is static and cannot adapt to student responses in real time. However, if we imagine substituting the AI avatar with an AI chatbot, we open the door to a written exam that can adapt dynamically to student input. This option is attractive, as it offers comparable benefits in scalability, anonymity, and analytical rigor, and may even prove more cost-effective. While technically feasible, this approach is not optimal given how the human brain processes information.^[307]

Summarizing a commonly used right-/left-brain framework as a heuristic, the FAA notes that written communication tends to favor analytic strengths, while real-time dialogue elicits contextual reasoning and adaptability.^[308] Those with left-brain dominance excel at writing and perform well on multiple-choice tests.^[309] They respond well to verbal instruction but often prefer structured tasks to spontaneous dialogue.^[310] By contrast, “those with right-brain dominance are characterized as being spatially oriented, creative, intuitive, and emotional.”^[311] Although the brain functions as a whole, most people exhibit a dominant side, and “when learning is new, difficult, or stressful, the brain seems to go on autopilot to the preferred side.”^[312]

Like aviation, real-world lawyering tasks also demand right-brain skills such as adaptability, intuition, and contextual understanding.^[313] As Paula Franzese observes, “much of what we tend to do in the law school classroom is aimed at honing left-brain thinking.”^[314] Right-brain functions are “more contextual,” encompassing “the abilities to glean patterns, perceive connections, craft meaningful narratives, navigate ethical quandaries, and understand the subtleties of human interaction.”^[315]

Franzese notes that these conceptual abilities are not easily automated, underscoring why oral exams are an effective method to verify authentic understanding.^[316] More importantly, these abilities are indispensable in the legal marketplace where, as Franzese predicted, “the left-brain functions are the ones most apt to get outsourced and automated.”^[317] Her observation, written more than a decade ago, was prescient given today’s rise of generative AI.^[318]

According to Franzese, “It behooves us to give those metaphorically ‘right hemisphere’ abilities—the more conceptual and contextual cognitive processes and skill sets—their due in the law school classroom.”^[319] Oral exams provide an avenue for right-brain dominant individuals to demonstrate strengths that written assessments may overlook. They also allow all students to develop and refine right-brain skills essential for effective lawyering. For these reasons, oral exams—whether administered by an AI avatar or a human examiner—should remain a core component of assessment. They uniquely call upon conceptual reasoning, judgment, adaptability, and metacognitive awareness, which are the capacities least susceptible to automation and most essential for competent lawyering.

No matter which model is used, legal writing faculty must ensure that the oral exam aligns with course objectives and learning outcomes. Questions should be carefully keyed to the skills and competencies the course is designed to develop. This alignment guarantees that the assessment measures what truly matters. In doing so, they also confirm that the student—not an algorithm—understands the analysis and concepts underlying their writing.

Major legal technology providers may also play an important role in advancing these assessment models. Companies that already support legal research technology are well-positioned to develop platforms capable of administering oral exams at scale. At a moment when generative AI integrations have prompted concerns about reduced cognitive engagement, tools designed to verify students’ reasoning rather than substitute for it would offer a constructive contribution. By supporting the administration of oral exams, these companies could strengthen academic integrity while helping law schools manage the logistical demands of a wider adoption of oral assessments.

Conclusion

At bottom, the rise of generative AI requires legal writing programs to adopt assessments that verify genuine understanding. Oral exams meet this need. Modeling an oral exam around a check-ride framework uses the same sequencing as aviation to verify mastery through oral questioning before a student demonstrates performance in a practical setting.^[320] This structure ensures that independent analytical skill is confirmed before students move on to simulation-based advocacy.

Several broader insights follow. Legal writing pedagogy can meaningfully draw on other disciplines, such as business, medicine, and software engineering, where instructors routinely integrate experiential and problem-centered learning.^[321] We can also learn from global traditions, including the longstanding use of oral exams that John Burman observed during his teaching in Russia.^[322] Looking to the past is equally instructive, as historical reliance on oral exams before the rise of mass education reminds us that forgotten methods may warrant revival when new challenges threaten the integrity of learning.^[323] Finally, like entrepreneurs, we should meet these challenges with principled innovation and adaptability.

Amidst all the uncertainty, human ingenuity remains the ultimate differentiator. If we apply these lessons, the solutions we create will prepare our students not only to succeed in a changing legal landscape but to lead with creativity, judgment, and resilience in a world where technology cannot replicate those qualities.^[324]

---

1. The term “generative AI” refers broadly to computerized tools that assist students in writing, including platforms like ChatGPT, Grammarly, and others that generate or suggest text. See generally Laurie Harris & Ling Zhu, Generative Artificial Intelligence and Data Privacy: A Primer, Cong. Rsch. Serv., R47569 (2023) (offering a clear explanation of how generative AI systems work); see also Sadie O’Connor, Generative AI, Geo. L. Tech. Rev. 394 (2024).

2. See, e.g., Andrew M. Perlman, Generative AI and the Future of Legal Scholarship (Mar. 3, 2024) (unpublished manuscript), https://ssrn.com/abstract=5072765 (arguing that generative AI can accelerate research and expand analytical capacity).

3. See, e.g., Anna C. Conley, Legal Education’s Role in Combating Automation Bias and Complacency with Generative AI (2024), https://scholarworks.umt.edu/faculty_lawreviews/228, Kilaw J. (forthcoming) (warning that AI reliance may erode foundational skills and professional judgment).

4. See, e.g., Amazon.com, Inc., Annual Report (Form 10-K) 25 (2026) (reporting $108.5 billion in technology and infrastructure expenses for fiscal year 2025); Alphabet Inc., Annual Report (Form 10-K) 35 (2026) (reporting $61.1 billion in research and development expenses for fiscal year 2025); Microsoft Corp., Annual Report (Form 10-K) 40 (2025) (reporting $32.5 billion in research and development expenses for fiscal year 2025); Meta Platforms, Inc., Annual Report (Form 10-K) 72 (2026) (reporting $57.4 billion in research and development expenses for fiscal year 2025).

5. See Kirsten K. Davis, A New Parlor is Open: Legal Writing Faculty Must Develop Scholarship on Generative AI and Legal Writing, 7 Stetson L. Rev. F. 1, 2 (2024) (stating that “generative AI is not going away, and it will likely impact how today’s law students complete the fundamental tasks of lawyering in their future practices”); see also Joe Regalia, From Briefs to Bytes: How Generative AI is Transforming Legal Writing and Practice, 59 Tulsa L. Rev. 193 (2024).

6. See, e.g., Nataliya Kosmyna, Eugene Hauptmann, Ye Tong Yuan, Jessica Situ, Xian-Hao Liao, Ashly Vivian Beresnitzky, Iris Braunstein & Pattie Maes, Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task, https://arxiv.org/pdf/2506.08872 [https://perma.cc/3DZ3-87RJ] (preprint, rev. Dec. 31, 2025); Samantha A. Moppett, Preparing Students for the Artificial Intelligence Era: The Crucial Role of Critical Thinking Skills, 52 Mitchell Hamline L. Rev. 240 (2025).

7. See Davis, supra note 5, at 20 (raising concerns about the reliability and authenticity of traditional assessments when AI can write for students).

8. See, e.g., Sidney Dekker & Judith Orasanu, Automation and Situation Awareness—Pushing the Research Frontier, in Coping with Computers in the Cockpit 69 (Sidney Dekker & Erik Hollnagel eds., 1999) (examining challenges posed by increasing cockpit automation); Charles Perrow, Normal Accidents: Living with High-Risk Technologies (1984) (analyzing how automation and system complexity shape accident dynamics).

9. See generally Fed. Aviation Admin., Airman Certification Standards—Private Pilot Airplane, FAA-S-ACS-6C (Apr. 2024, current for 2026) (providing standardized training expectations, including risk-management and judgment components); see also David D. Woods & Erik Hollnagel, Joint Cognitive Systems: Patterns in Cognitive Systems Engineering (2006) (outlining training methods designed to prepare humans to work effectively with automated systems).

10. See Fed. Aviation Admin., supra note 9 (detailing the oral knowledge evaluation integral to the check-ride).

11. Admittedly, their complexity might make AI-assisted assessment resemble the automobile industry’s development of automated-driving systems more. Automated-driving technologies rely on dynamic sensor environments and continuous data processing, much like the multifaceted inputs an AI oral exam platform would need to evaluate student reasoning in real time. This is not to downplay the sophistication of autopilot systems; rather, the point is that autopilot operates on a comparatively narrow band of inputs (such as altitude and heading), whereas automated-driving systems and AI-supported assessment must interpret a far broader and more fluid stream of information.

12. See Am. Bar Ass’n, ABA Standards and Rules of Procedure for Approval of Law Schools Stand. 314 (2025–2026), https://www.americanbar.org/content/dam/aba/administrative/legal_education_and_admissions_to_the_bar/standards/2025-2026/2025-2026-standards-and-rules-of-procedure-for-approval-of-law-schools.pdf [hereinafter ABA Standards] (requiring law schools to provide both formative and summative assessment techniques to measure and improve student learning).

13. See id. Stand. 303(b)(3) (requiring law schools to “provide substantial opportunities” for students to develop a professional identity).

14. This Article merely notes the model’s potential adaptability to other instructional settings; it does not attempt to develop or assess those applications.

15. See Mark Cuban Discusses Artificial Intelligence at National Governors Association, C-SPAN (July 25, 2025), https://www.c-span.org/program/public-affairs-event/mark-cuban-discusses-artificial-intelligence-at-national-governors-association/663103 [hereinafter Cuban Interview].

16. See id.

17. See Self-licking ice cream cone, Wikipedia, https://en.wikipedia.org/wiki/Self-licking_ice_cream_cone (last visited Jan. 19, 2026) (summarizing origin and modern uses in public policy contexts).

18. As Cuban put it, “There are going to be two types of companies in this world: those who are great at AI and everybody else.” Cuban Interview, supra note 15. Likewise, there will be two types of legal educators: those who excel at teaching in an environment increasingly dependent on AI and everyone else.

19. Jan Levine has observed that progress in legal writing pedagogy depends on maintaining continuity with the field’s accumulated experience. See Jan M. Levine, A Curmudgeon’s View of the Multi-Generational Teaching of Legal Writing, 25 Legal Writing 79, 80–81 (2021). His reflections underscore the importance of approaching new challenges, such as the rise of generative AI, with both innovation and an appreciation for established principles.

20. See generally Kristen K. Robbins-Tiscione, A Call to Combine Rhetorical Theory and Practice in the Legal Writing Classroom, 50 Washburn L.J. 319 (2011); see also Linda L. Berger, Studying and Teaching “Law as Rhetoric”: A Place to Stand, 16 Legal Writing 3, 11 (2010). As one scholar observes, viewing “law as rhetoric is essential to begin the complex task of legal interpretation,” and rhetoric “is essential for legal composition, perhaps even more naturally so because rhetoric is the historical site of the tools and implements of persuasion and argumentation.” Berger, supra, at 11. Aristotle’s five canons (invention, arrangement, style, memory, and delivery) and the modes of persuasion (ethos, pathos, logos) remain foundational. See Robbins-Tiscione, supra, at 325; see also Michael R. Smith, Rhetoric Theory and Legal Writing: An Annotated Bibliography, 3 J. Ass’n Legal Writing Directors 129, 130 (2008).

21. Robbins-Tiscione, supra note 20, at 321. Marjorie Rombauer, widely regarded as the founder of the legal writing field, traced the development of legal writing courses from the early 1900s through the 1970s. See generally Marjorie D. Rombauer, First-Year Legal Research and Writing: Then and Now, 25 J. Legal Educ. 538 (1973).

22. Robbins-Tiscione, supra note 20, at 321.

23. Id.

24. Id. at 324 (citing Rombauer, supra note 21, at 540).

25. See Am. Bar. Ass’n, Sourcebook on Legal Writing Programs 208 (J. Lyn Entrikin ed., 3d ed. 2020) [hereinafter Sourcebook 3d ed.]; see also An Interview with Legal Writing Professor Jan Levine, WordRake Blog, https://www.wordrake.com/blog/an-interview-with-legal-writing-professor-jan-levine [https://perma.cc/XPH6-T9CB] (last visited Apr. 2, 2026) [hereinafter Levine Interview].

26. See Mary Beth Beazley, Better Writing, Better Thinking: Using Legal Writing Pedagogy in the “Casebook” Classroom (Without Grading Papers), 10 Legal Writing 23, 25–27 (2004).

27. See generally Robert J. Condlin, “Practice Ready Graduates”: A Millennialist Fantasy, 31 Touro L. Rev. 75 (2015); Lisa T. McElroy, Christine N. Coughlin & Deborah S. Gordon, The Carnegie Report and Legal Writing: Does the Report Go Far Enough?, 17 Legal Writing 279 (2011). Skeptics have criticized the notion of “practice-ready,” much of which stems from the Carnegie Report. See Condlin, supra, at 96–102. The Carnegie Report, a landmark study on legal education, “stirred the pot” by emphasizing integration of practical skills with doctrinal learning. Id. at 103–04 n. 76; see generally William M. Sullivan, Anne Colby, Judith Welch Wegner, Lloyd Bond & Lee S. Shulman, Educating Lawyers: Preparation for the Profession of Law (2007) (commonly known as the Carnegie Report). Nevertheless, it praised legal writing professors for embracing interactive and experiential teaching methods. See McElroy et al., supra, at 287–89.

28. See generally L. Danielle Tully, Behind the Curve: Rethinking Norm-Referenced Grading in First-Year Legal Writing Courses, 29 Legal Writing 1 (2025). Empirical research shows that students who succeed in legal writing often succeed across the curriculum and in employment outcomes. Jessica L. Clark, Grades Matter; Legal Writing Grades Matter Most, 32 Miss. C. L. Rev. 375, 413–18 (2014). The Legal Writing Sourcebook offers a leading account of grading in legal writing courses, outlining approaches designed to promote transparency, consistency, and alignment with articulated learning objectives. See Sourcebook 3d ed., supra note 25, at 206–09.

29. Clark, supra note 28, at 376 (noting literature recognizing the central importance of grades to students and employers).

30. See, e.g., Jessica L. Clark & Christy Hallam DeSanctis, Toward a Unified Grading Vocabulary: Using Rubrics in Legal Writing Courses, 63 J. Legal Educ. 1 (2013).

31. See, e.g., Deborah L. Borman, De-grading Assessment: Rejecting Rubrics in Favor of Authentic Analysis, 41 Seattle U. L. Rev. 713 (2018).

32. See Davis, supra note 5, at 21 (questioning whether grading finished legal writing remains an effective way to assess student skills in the age of generative AI).

33. See generally id.; see also John Bliss, Teaching Law in the Age of Generative AI, 64 Jurimetrics J. 111 (2024).

34. See Bliss, supra note 33, at 127–31.

35. Id. at 114.

36. Id. To address these challenges, Bliss suggests that “[o]ral exams, as practiced in many European universities, can provide the instructor a direct opportunity to inquire about each student’s depth of knowledge, understanding, and ability.” Id. at 160.

37. Id. at 127–31.

38. See Stephanie Wilkins, Orrick Trains Summer Associates in Prompt Engineering with New Course from AltaClaro, Law​.​com (Aug. 9, 2023, 12:20 PM), https://www.law.com/therecorder/2023/08/09/orrick-trains-summer-associates-in-prompt-engineering-with-new-course-from-altaclaro [https://perma.cc/U37L-WZM3].

39. Bliss, supra note 33, at 138. Both faculty and student respondents to the survey agreed that students must develop AI literacy. Id.

40. Id. (noting that “[p]reparing practice-ready graduates requires not only instructing students in the proficient use of AI tools but also instilling in them a commitment to use these tools responsibly”).

41. See Davis, supra note 5, at 2–5.

42. Id. (arguing that generative AI marks “the start of a paradigm shift in legal writing” requiring focused scholarly attention).

43. Id. at 20 (emphasis in original).

44. Id.

45. See generally Carolyn V. Williams, Bracing for Impact: Revising Legal Writing Assessments Ahead of the Collision of Generative AI and the NextGen Bar Exam, 28 Legal Writing 1 (2024).

46. See id. at 1–9.

47. Id. at 34–46.

48. Id. at 56.

49. See Bliss, supra note 33, at 130 (finding that, although most faculty agreed that generative AI should be incorporated into legal education, many also worried that AI could “‘hinder’ the teaching of law and ‘compromise the learning process’”).

50. Davis, supra note 5, at 21 (observing that generative AI may interfere with the ability of faculty to rely on written work as indirect evidence of learning and suggesting that legal writing professors may need to adopt alternative assessments).

51. Id. at 20 (discussing the ease with which students may improperly rely on generative AI tools in ways that violate assignment or institutional policies).

52. Id. at 10–11 (explaining that generative AI unsettles traditional assumptions about authorial intention, authenticity, and responsibility, requiring scholars to “reconceptualize what it means to be an ‘author’” of legal writing).

53. Id. (discussing questions about authenticity and authorship in legal writing). While authorship concerns are important, more relevant to legal writing instruction is whether a student’s work shows genuine understanding. Before the advent of generative AI, students often produced similar work from uniform assignments. Many of them also received support from teaching assistants or writing specialists, none of whom are considered co-authors. See Patricia Grande Montana, A Contemporary Model for Using Teaching Assistants in Legal Writing Programs, 42 Mitchell Hamline L. Rev. 185, 186–91 (2016) (describing teaching assistants’ roles in instruction, feedback, and mentoring within legal writing programs); Julie M. Cheslik, Teaching Assistants: A Study of Their Use in Law School Research and Writing Programs, 44 J. Legal Educ. 394, 395–401 (1994) (surveying teaching assistant functions). The main point is whether students engaged with the material with strong legal research and reasoning.

54. See Bliss, supra note 33, at 129–30 (reporting faculty concerns about students’ overreliance on AI tools and emphasizing the need for students to understand the risks and limitations associated with legal AI systems).

55. Williams, supra note 45, at 23–26 (discussing the persistent weaknesses of generative AI tools).

56. Id. at 23.

57. Bliss, supra note 33, at 141–42.

58. Williams, supra note 45, at 1–2 (illustrating, through a hypothetical, how a student who relies heavily on generative AI may appear to meet learning outcomes despite insufficient engagement with the writing process); see also Mary Burns, Rebecca Winthrop, Natasha Luther, Emma Venetis & Rida Karim, A New Direction for Students in an AI World: Prosper, Prepare, Protect, Brookings Institution (Jan. 14, 2024), https://www.brookings.edu/articles/a-new-direction-for-students-in-an-ai-world-prosper-prepare-protect [https://perma.cc/523Y-R2AQ] (documenting reduced attention and skill development associated with heavy generative AI reliance); Hao-Ping (Hank) Lee, Advait Sarkar, Lev Tankelvitch, Ian Drosos, Sean Rintel, Richard Banks & Nicholas Wilson, The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence—Effects From a Survey of Knowledge Workers, Microsoft Research (2025), https://www.microsoft.com/en-us/research/wp-content/uploads/2025/01/lee_2025_ai_critical_thinking_survey.pdf [https://perma.cc/TTB2-AU4H] (finding that routine AI use can diminish cognitive effort).

59. Bliss, supra note 33, at 130 (quoting a faculty respondent emphasizing that students must “learn the hard way” and master basic legal skills before using AI tools as a supplement).

60. Levine Interview, supra note 25 (noting that legal writing courses give students “a concrete and hands-on introduction to a new discourse community”).

61. Cameron R. Jones & Benjamin K. Bergen, Large Language Models Pass the Turing Test, https://arxiv.org/pdf/2503.23674 [https://perma.cc/8NDU-KJHW] (preprint, Mar. 31, 2025); see generally A. M. Turing, Computing Machinery and Intelligence, 59 Mind 433 (1950); see also Adam A. Bent, Large Language Models: AI’s Legal Revolution, 44 Pace L. Rev. 91, 101 (2023) (explaining how Turing claimed “[a] computer exhibits intelligence if it can deceive a person into believing that the computer is human”); Deborah W. Denno & Erica Valencia-Graham, Symposium: The New AI: The Legal and Ethical Implications of ChatGPT and Other Emerging Technologies: Forward, 92 Fordham L. Rev. 1785, 1786 (2024) (noting that “Turing’s thoughts and predictions were as revelatory as they were startling”).

62. See Turing, supra note 61. One anthropology expert explains that Turing “considered an intelligent computing machine to be rendered more human by virtue of thinking like a person, rather than looking like one. This perspective has been received and integrated into the foundational theories of contemporary AI.” Ting Guo, Alan Turing: Artificial Intelligence as Human Self-Knowledge, 31 Anthropology Today, No. 6, at 3 (2015).

63. Turing, supra note 61, at 433 (“I propose to consider the question, ‘Can machines think?’”).

64. Turing began his analysis “with definitions of the meaning of the terms ‘machine’ and ‘think,’” cautioning that these “definitions might be framed so as to reflect so far as possible the normal use of the words, but this attitude is dangerous . . . .” Id.; see also Peter Bieri, Thinking Machines: Some Reflections on the Turing Test, 9 Poetics Today 163, 164 (1988) (explaining that Turing aimed to avoid sterile semantic debates).

65. Turing, supra note 61, at 433. Turing observed that “it is difficult to escape the conclusion that the meaning and the answer to the question, ‘Can machines think?’ is to be sought in a statistical survey such as a Gallup poll. But this is absurd.” Id.

66. See id. Turing explained, “Instead of attempting such a definition I shall replace the question by another, which is closely related to it and is expressed in relatively unambiguous words.” Id.

67. Turing, supra note 61, at 433; see also Jones & Bergen, supra note 61, at 1 (describing the Turing test).

68. Jones & Bergen, supra note 61, at 1.

69. Id.

70. Id.

71. Id.

72. Id. at 2 (“Over the last 75 years there have been many attempts to construct systems that could pass the Turing test”).

73. LLMs are a form of generative AI that involve “connectionist systems which learn to produce language on the basis of distributional statistics and reinforcement learning feedback.” Id.

74. Id. (“[N]one have succeeded”).

75. Id. at 8, 12. The study also used a fourth AI model, ELIZA, described as “a rules-based chatbot from the 1960s.” Id. at 2–3. The results did not indicate that ELIZA was reliably indistinguishable from a human. Id. at 8.

76. Id.

77. Daniel J. Solove & Hideyuki Matsumi, AI, Algorithms, and Awful Humans, 92 Fordham L. Rev. 1923, 1925 (2024) (“Machine decision-making currently cannot incorporate emotion, morality, or value judgments, which are essential components of decisions involving people’s welfare”).

78. E.g., Model Rules of Pro. Conduct r. 1.1 (A.B.A. 2020) (“A lawyer shall provide competent representation to a client. Competent representation requires the legal knowledge, skill, thoroughness and preparation reasonably necessary for the representation.”); id. r. 3.3 (“A lawyer shall not knowingly . . . make a false statement of fact or law to a tribunal or fail to correct a false statement of material fact or law previously made”).

79. Bent, supra note 61, at 101. In assessing the impact of AI on law, Adam Bent remarked, “The Turing Test was the catalyst for the chatbot landscape we face today, leaving lawyers, judges, bar associations, law schools, and society at large scrambling to grapple with the implications of such advanced technology.” Id.

80. Id. at 94–96 (discussing how LLMs are becoming exceedingly fast and accurate for a plethora of applications).

81. Daniel Solove and Hideyuki Matsumi, leading voices in law, privacy, and technology, observe that AI is being used “in a myriad of decisions about people’s freedom, opportunities, and welfare. Algorithms are deployed in decisions as diverse as hiring, criminal sentencing, education, and lending.” Solove & Matsumi, supra note 77, at 1924. Their research frames the debate through two contrasting perspectives: the “Awful Human Argument” and the “Better Together Argument.” Id. The former “asserts that human decision-making is often bad and that machines can decide better than humans.” Id. The later “posits that machines can augment and improve human decision-making.” Id.

82. See generally Madeleine C. Elish & Tim Hwang, Praise the Machine! Punish the Human! The Contradictory History of Accountability in Automated Aviation, Working Paper No. 1, Intelligence & Autonomy Initiative, Data & Society Research Institute (Feb. 24, 2015), ), https://datasociety.net/wp-content/uploads/2026/03/Elish-Hwang_AccountabilityAutomatedAviation.pdf [https://perma.cc/7WLZ-QAAY]; see also Pia Bergqvist, Flight School: Teaching Automation, Flying Mag. (July 23, 2011), https://www.flyingmag.com/training-learn-fly-flight-school-teaching-automation [https://perma.cc/L4L2-49YR].

83. Elish & Hwang, supra note 82, at 6 (“As technological advances increased the safety and quantity of commercial aircraft, laws and social norms shifted to meet the new technologies”); see generally Raja Parasuraman, Thomas B. Sheridan & Christopher D. Wickens, A Model for Types and Levels of Human Interaction with Automation, 30 IEEE Transactions on Sys., Man & Cybernetics —Part A: Systems and Humans No. 3, 286 (2000) (suggesting a framework for categorizing levels of human interaction with automation, using air traffic control systems as an example).

84. Elish & Hwang, supra note 82, at 4. “The first flight to successfully use Sperry’s autopilot took place in New York in 1913.” Id. at 4 n.7; see also Luke Peters, Examining over 100 years of flight automation and the history of the autopilot, AeroTime (Apr. 4, 2025), https://www.aerotime.aero/articles/autopilot-flight-automation-history [https://perma.cc/42EU-VYQ8].

85. Elish & Hwang, supra note 82, at 4 n.8.

86. See id. Unlike the automated features found in self-driving cars, which are designed to reduce or even remove the need for human oversight, aviation autopilots assume the presence of an engaged pilot who remains responsible for monitoring, judgment, and intervention. And unlike automated-driving features, which are engineered so that ordinary drivers can use them without extensive instruction, aviation autopilot presupposes a pilot who has undergone rigorous training and who remains responsible for continuous oversight and decision-making. See, e.g., Fed. Aviation Admin, Pilot’s Handbook of Aeronautical Knowledge at ch. 2, 25-2–2-32 (2023), https://www.faa.gov/regulations_policies/handbooks_manuals/aviation/faa-h-8083-25c.pdf [https://perma.cc/GEJ3-RKXA] (explaining that autopilot systems assist but do not replace pilot judgment or responsibility); SAE Int’l, J3016_202104 Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles (Apr. 29, 2021), https://www.sae.org/standards/j3016_202104-taxonomy-definitions-terms-related-driving-automation-systems-road-motor-vehicles [https://perma.cc/DHC3-U675] (defining vehicle automation levels ranging from no driving automation (Level 0) to full driving automation (Level 5)).

87. See Aviation Tech: History of Autopilot and Its Role in Modern Private Flight, SHY Aviation (July 7, 2025), https://www.shyaviation.com/blog/aviation-tech-history-of-autopilot-and-its-role-in-modern-private-flight [https://perma.cc/CS6A-N8NL].

88. Elish & Hwang, supra note 82, at 4 n.8.

89. For purposes of the present discussion, the term “autopilot” is used broadly and does not refer to any specific technical configuration. It serves as a conceptual analogue to generative AI in the legal profession.

90. Bergqvist, supra note 82.

91. See LaQuinton Armbrister, Automation in Aviation: An Advancement or Hindrance to Aviation Safety? (Master’s thesis, Tex. S. Univ. 2023), https://digitalscholarship.tsu.edu/cgi/viewcontent.cgi?article=1067&context=theses (discussing challenges of automation in aviation).

92. Fed. Aviation Admin., CFIT/Automation Overreliance (Feb. 4, 2022), https://www.faa.gov/sites/faa.gov/files/2022-01/CFIT Automation Overreliance.pdf [https://perma.cc/LQ7E-PYMY]. The General Aviation Joint Safety Committee has identified overreliance on automation as a key contributor to controlled flight into terrain incidents, due to its tendency to foster complacency and skill erosion. CFIT and Overreliance on Automation, FAA Safety Briefing Mag. (Nov. 2021), https://medium.com/faa/cfit-and-overreliance-on-automation-48eed783b1e9; see also Jason Blair, The Dangers of Overreliance on Automation, FAA Safety Briefing Mag. (May/June 2025), https://jasonblair.net/wp-content/uploads/2025/05/Jason-Blair-FAA-Safety-Briefing-MayJun2025-The-Dangers-of-Overreliance-on-Automation-Safety-Concerns-and-Mitigation-Strategies-for-Pilots.pdf [https://perma.cc/C7KF-GQTZ].

93. Fed. Aviation Admin., supra note 92.

94. Antonio Chialastri, Automation in Aviation, in Automation 79, 94–95 (Florian Kongoli ed., 2012).

95. Fed. Aviation Admin., supra note 92.

96. Chialastri, supra note 94, at 95 (“When automation fails or behaves in a ‘strange’ manner, the workload increases exponentially”).

97. Fed. Aviation Admin., Summary of the FAA’s Review of the Boeing 737 MAX 5 (Nov. 18, 2020), https://www.faa.gov/sites/faa.gov/files/2022-08/737_RTS_Summary.pdf [https://perma.cc/XW25-SCSE]; Bart Elias, Rachel Y. Tang, Daniel Morgan, Laura B. Comay, Richard K. Lattanzio, Kelsi Bracmort & Martin Offutt, Federal Aviation Administration (FAA) Reauthorization Issues for the 118th Congress, Cong. Rsch. Serv., R47608 (2023).

98. Fed. Aviation Admin., U.S. Dep’t of the Air Force, NASA & U.S. Dep’t of Transp., Technical Advisory Board on the Design Change to the B737 MAX Maneuvering Characteristics Augmentation System ii (Nov. 18, 2020), https://www.faa.gov/sites/faa.gov/files/2022-08/737_Technical_Advisory_Board_Final_Report.pdf [https://perma.cc/JK4E-6P9A].

99. See Fed. Aviation Admin., supra note 97; see also Nat’l Transp. Safety Bd., Response to Final Aircraft Accident Investigation Report: Ethiopian Airlines Flight 302, 1 n.1 (Jan. 13, 2023), https://www.ntsb.gov/investigations/Documents/Response to EAIB final report.pdf [https://perma.cc/64RJ-S7UU].

100. Nat’l Transp. Safety Bd., supra note 99, at 4–5.

101. Dominic Gates & Mike Baker, The Inside Story of MCAS: How Boeing’s 737 MAX System Gained Power and Lost Safeguards, Seattle Times (June 22, 2019), https://www.seattletimes.com/seattle-news/times-watchdog/the-inside-story-of-mcas-how-boeings-737-max-system-gained-power-and-lost-safeguards [https://perma.cc/2ZZN-JRS4].

102. See Fed. Aviation Admin., supra note 97.

103. See Armbrister, supra note 91, at 1 (“Automation’s presence in the cockpit has been quite advantageous by improving economics, enhancing safety, and arguably reducing workload”).

104. See id. at 5–9 (discussing the advantages of automation in aviation); see also Rachel Gordon, AI Copilot Enhances Human Precision for Safer Aviation, MIT News (Oct. 3, 2023), https://news.mit.edu/2023/ai-co-pilot-enhances-human-precision-safer-aviation-1003 [https://perma.cc/H4P7-NA8Y] (explaining how advanced autopilot systems function).

105. See Armbrister, supra note 91, at 7–8 (discussing the “importance of managing workload” by pilots). Similar dynamics appear across other fields that rely on high-stakes and time-sensitive decision-making. For example, computer systems can now assist in decisions traditionally made exclusively by human experts, such as whether an anesthesiologist should adjust a patient’s medication, whether an air defense operator should engage an approaching aircraft, or whether a securities analyst should purchase a large block of stock. Parasuraman et al., supra note 83, at 286.

106. Bergqvist, supra note 82.

107. Amanda Grondin, Effectiveness of the Socratic Method: A Comparative Analysis of the Historical and Modern Invocations of an Educational Method, 1, 4 (2018), https://scholarcommons.sc.edu/cgi/viewcontent.cgi?article=1254&context=senior_theses (unpublished S.M. thesis, Univ. of S. Carolina).

108. See Elizabeth A. Usman, Making Legal Education Stick: Using Cognitive Science to Foster Long-Term Learning in the Legal Writing Classroom, 29 Geo. J. Legal Ethics, 355, 357 (2016) (examining recent trends in which legal scholars reconsider traditional teaching methods in light of cognitive science research).

109. U.S. Dep’t of Transp. & Fed. Aviation Admin., Aviation Instructor’s Handbook at ch. 3, 3-2–3-4 (2020), https://www.faa.gov/sites/faa.gov/files/regulations_policies/handbooks_manuals/aviation/aviation_instructors_handbook/aviation_instructors_handbook.pdf [hereinafter Aviation Instructor’s Handbook] (defining “learning theory” and explaining that cognitive theory concerns internal mental processes).

110. Usman, supra note 108, at 381–82.

111. See Aviation Instructor’s Handbook, supra note 109, at ch. 3, 3-2–3-7.

112. Id. at Preface, ii.

113. Id.

114. Id. at ch. 3, 3-7.

115. Id. at ch. 3, 3-7–3-8.

116. Id. at ch. 3, 3-9 (explaining that learners acquire “insight” by grouping perceptions into meaningful patterns and that instruction accelerates this process through structured guidance).

117. Id.

118. Id. at ch. 3, 3-5.

119. Id. at ch. 3, 3-14; see generally Comm. of Coll. & Univ. Exam’rs, Taxonomy of Educational Objectives: The Classification of Educational Goals, Handbook I: Cognitive Domain (Benjamin S. Bloom ed., 1956).

120. See Aviation Instructor’s Handbook, supra note 109, at ch. 3, 3-14.

121. See id. These distinctions map naturally onto legal writing skills. A student who can state the elements of negligence demonstrates Knowledge, while a student who can analyze whether a court would recognize a new duty in a novel factual situation is operating at the level of Evaluation.

122. See id. at ch. 3, 3-5–3-6. The Aviation Instructor’s Handbook also explains behaviorism as a learning theory. Id. at ch. 3, 3-3 (“Behaviorism explains animal and human behavior entirely in terms of observable and measurable responses to stimuli”). Behaviorism is the least relevant to the present discussion; behaviorist models do not account for the critical thinking and metacognitive skills central to legal writing instruction. See id.

123. Id. at ch. 3, 3-5. The Aviation Instructor’s Handbook explains that “[a] computer gets input from a keyboard, mouse, etc., whereas the human brain gets input from the senses of sight, hearing, touch, taste, and smell.” Id.

124. See id.

125. Id. at ch. 3, 3-5. To make this point, the Aviation Instructor’s Handbook states that “[t]he amount of sensory input the brain receives per second ranges from thousands to millions of bits of information according to various theories. Regardless of the number, that is a lot of information for the brain to track and process.” Id. In the context of aviation pedagogy, this aspect is critical because cognitive overload can impair learning. Just like aviation instructors must design training to avoid cognitive overload, legal writing professors should too design courses to account for information processing strain.

126. Id. at ch. 3, 3-5–3-6.

127. Id. at ch. 3, 3-5–3-7.

128. Id. at ch. 3, 3-5.

129. Id. at ch. 3, 3-6.

130. Id.

131. Id.

132. Id. at ch. 3, 3-6, 3-42.

133. Id. at ch. 3, 3-6—3-7 (noting that scenario-based training integrates aeronautical knowledge with decision-making through active, contextual participation).

134. Id. at ch. 5, 5-17.

135. Id. at ch. 3, 3-6. According to the Aviation Instructor’s Handbook, “It is being used to train people in everything from emergency response to hotel management.” Id.

136. See id. at ch. 3, 3-5—3-6.

137. See id.; see also K. Anders Ericsson, Ralf Th. Krampe & Clemens Tesch-Romer, The Role of Deliberate Practice in the Acquisition of Expert Performance, 100 Psychol. Rev. 363, 367–69 (1993) (introducing the theory of “deliberate practice” and concluding that expertise develops through structured practice with immediate feedback).

138. See Aviation Instructor’s Handbook, supra note 109, at ch. 6, 6-1.

139. Id. at ch. 6, 6-3.

140. Id. at ch. 6, 6-5.

141. Id.

142. See 14 C.F.R. §§ 61.87(d), 61.103(d)-(e) (requiring student pilots to complete ground training, receive instructor endorsements, and demonstrate proficiency in specified maneuvers before being eligible for solo flight, the FAA knowledge test, or the FAA practical test).

143. See 14 C.F.R. § 61.87(d) (listing required pre-solo maneuvers and procedures); U.S. Dep’t of Transp. & Fed. Aviation Admin., Advisory Circular 61-65H, Certification: Pilots and Flight and Ground Instructors 8 (Aug. 27, 2018), https://www.faa.gov/documentLibrary/media/Advisory_Circular/AC_61-65H.pdf [https://perma.cc/B7C4-4VRU] (explaining prerequisites for knowledge and practical tests, including ground training, instructor endorsements, and completion of required flight training).

144. See 14 C.F.R. § 61.43(a) (requiring examiners to evaluate both knowledge and flight proficiency during the practical test as specified in the Airman Certification Standards). Commercial aviation also relies on oral exams as part of its ongoing competency standards. See 14 C.F.R. § 135.293(a).

145. Aviation Instructor’s Handbook, supra note 109, at ch. 6, 6-3–6-5 (discussing aspects of effective traditional assessment).

146. Id. at ch. 6, 6-5–6-7 (discussing aspects of effective authentic assessment).

147. 14 C.F.R. § 61.43(a).

148. See Fed. Aviation Admin., U.S. Civil Airmen Statistics, Calendar Year 2024, tbls. 19, 20 (2025), https://www.faa.gov/data_research/aviation_data_statistics/civil_airmen_statistics (click “2025 Active Civil Airmen Statistics” under “Annual Statistics”). In 2025, examiner-conducted original private pilot practical tests had an approval rate of 75.1 percent, yielding a disapproval rate of 24.9 percent. Id. tbl. 19. The overall approval rate across all original pilot certificate categories was 78.4 percent. Id. Pass rates for additional ratings (sought by certificated pilots adding new aircraft categories or instrument qualifications) were notably higher, averaging 87.8 percent, reflecting the greater experience of those applicants. Id. tbl. 20.

149. See generally Aviation Instructor’s Handbook, supra note 109, at ch. 1 (discussing the importance of communication, stress adaptation, situational awareness, and detail-oriented monitoring for pilots); Am. Bar Ass’n Section of Legal Educ. & Admissions to the Bar, Legal Education and Professional Development—an Educational Continuum (1992) (identifying fundamental lawyering skills and emphasizing that legal education must develop these competencies through realistic performance settings). The ABA report is known as the MacCrate Report after the Task Force’s Chairperson Robert MacCrate.

150. Davis, supra note 5, at 2 (citing Sarah E. Eaton, 6 Tenets of Post Plagiarism: Writing in the Age of Artificial Intelligence, Medium (Feb. 25, 2023), https://medium.com/@sarahelaineeaton/six-tenets-of-postplagarism-writing-in-the-age-of-artificial-intelligence-6340b809cba4); see also Williams, supra note 45, at 21–23 (describing the general strengths of AI tools).

151. See Davis, supra note 5, at 9 (discussing how generative AI may integrate into the legal writing workflow); see also Bliss, supra note 33, at 151–52 (“[S]tudents who learn effective prompt engineering techniques may be able to enhance the efficiency and quality of their legal work”).

152. Davis, supra note 5, at 9 (emphasis in original).

153. Bliss, supra note 33, at 139.

154. See Gina J. Mariano, Debra E. Allwardt, Paul R. Raptis & Kristine Stillwell, Reintroducing the Oral Exam: Finding Out What Your Students Really Know in the Age of ChatGPT, Currents in Teaching & Learning 59, 59 (Sep. 2024) (observing that generative AI allows students to complete assignments “with almost no skill or effort required”).

155. See, e.g., Benjamin Weiser, ChatGPT Lawyers Are Ordered to Consider Seeking Forgiveness, N.Y. Times (June 22, 2023), https://www.nytimes.com/2023/06/22/nyregion/lawyers-chatgpt-schwartz-loduca.html; Sara Merken, New York Lawyers Sanctioned for Using Fake ChatGPT Cases in Legal Brief, Reuters (June 26, 2023), https://www.reuters.com/legal/new-york-lawyers-sanctioned-using-fake-chatgpt-cases-legal-brief-2023-06-22.

156. See Damien Charlotin, Hallucinations: A Tracker of AI-Generated Fabrications, https://www.damiencharlotin.com/hallucinations [https://perma.cc/3DS8-NWN7] (last visited July 7, 2026) (tracking AI-generated hallucinations). Legislative responses are also emerging. Emily Sawicki, Calif. Bill Would Ban AI from Replacing Arbitrators’ Analysis, Law360 (Jan. 14, 2026), https://www.law360.com/legalethics/articles/2430003. However, these concerns are likely short-term challenges and should not discourage legal writing professors from teaching students how to use generative AI tools effectively.

157. Mariano et al., supra note 154, at 59.

158. Id.

159. Id.; see also Am. Bar Ass’n, Report to the House of Delegates, Resolution 112 (Aug. 12-13, 2019), https://www.americanbar.org/content/dam/aba/directories/policy/annual-2019/112-annual-2019.pdf (last visited Mar. 20, 2026) (urging lawyers and law schools to recognize technological competence as a core professional obligation).

160. ABA Standards, supra note 12, Stand. 314.

161. Id. Interpretation 314-1.

162. Aviation Instructor’s Handbook, supra note 109, at ch. 6, 6-1.

163. Id.

164. ABA Standards, supra note 12, Interpretation 314-1.

165. Id.

166. Id. Stand. 303(b)(3).

167. Id. Interpretation 303-5.

168. See Bliss, supra note 33, at 140 (observing that professionals in multiple fields “have reported feeling dejected and somewhat mystified when AI accomplishes much of what they understand to be their craft”) (citing Josie Cox, AI Anxiety: The Workers Who Fear Losing Their Jobs to Artificial Intelligence, BBC (July 13, 2023), https://www.bbc.com/worklife/article/20230418-ai-anxiety-artificial-intelligence-replace-jobs [https://perma.cc/A4ED-3AFJ]).

169. E. Scott Fruehwald, Developing Law Students’ Professional Identities, 37 U. La Verne L. Rev. 1, 5 (2015) (citing Maria Cardelle-Elawar, Leslie Irwin & Maria Luisa Sanz de Acedo Lizarraga, A Cross Cultural Analysis of Motivational Factors that Influence Teacher Identity, 5 Elec. J. Res. Educ. Psych. 565, 569–70 (2007)).

170. Fruehwald, supra note 169, at 5–7 (citing Michael Hunter Schwartz, Teaching Law by Design: How Learning Theory and Instructional Design Can Inform and Reform Law Teaching, 38 San Diego L. Rev. 347, 376 (2001)).

171. See generally id. at 7–14 (citing Robin Wellford Slocum, An Inconvenient Truth: The Need to Educate Emotionally Competent Lawyers, 45 Creighton L. Rev. 827, 829 (2012)).

172. Id. at 14–15; see also Donald A. Schön, The Reflective Practitioner: How Professionals Think in Action 49–50 (1983) (explaining that professionals develop judgment through “reflection-in-action,” a process that occurs in real time while responding to unfolding circumstances).

173. See, e.g., Fed. Aviation Admin., supra note 86, at ch. 2, 2-26 (explaining that “[a]utomation is the single most important advance in aviation technologies”); see generally Melissa H. Weresh, I’ll Start Walking Your Way, You Start Walking Mine: Sociological Perspectives on Professional Identity Development and Influence of Generational Differences, 61 S.C. L. Rev. 337, 340–45 (2009) (examining the role of generational differences in professional identity formation, including how technology and instant access to information influence approaches to legal education and practice).

174. See Bliss, supra note 33, at 124–26 (describing students’ initial awe and intimidation of perceived AI capabilities). Post-Watergate professionalism reform sought to address concerns that lawyers had relinquished individual responsibility for judgment in favor of external authority; similar concerns arise today when professional judgment is deferred to algorithmic outputs. Cf. Alison Donahue Kehner & Mary Ann Robinson, Mission: Impossible, Mission: Accomplished or Mission: Underway? A Survey and Analysis of Current Trends in Professionalism Education in American Law Schools, 38 U. Dayton L. Rev. 57, 62–63 (2012).

175. See generally supra Part II. The reader might find a more familiar analogy in the development of DNA technology. Investigators gained unprecedented evidentiary resources, yet their interpretive and ethical responsibilities actually deepened. The contrast between Sherlock Holmes at Baker Street, inferring meaning from a thread of cigar ash, and the lab-driven world of the television series NCIS captures the same dynamic. The technological advances reshape the tools available to a profession but not the core judgment on which professional identity rests.

176. See supra Part III.B.

177. See Kehner & Robinson, supra note 174, at 63–67 (citing the MacCrate Report, supra note 149, and the Carnegie Report, supra note 27).

178. James Naughton, Testocracy: The Undemocratic System of Standardized Testing in the United States, 31 Kan. J.L. & Pub. Pol’y 263, 290–91 (2022). Along with essays, oral exams were “considered the most rigorous and valid test of academic competence.” Id. at 265 (citing Doug A. Archbald & Fred M. Newmann, Beyond Standardized Testing: Assessing Authentic Achievement in Secondary School 7, 22 (1988)).

179. Off. Tech. Assessment, U.S. Cong. OTA-SET-519, Testing in American Schools: Asking the Right Questions 104 (1992) (explaining that written tests replaced oral exams as schools shifted from elite education to mass education).

180. Id. at 106–07 (explaining the adoption of standardized tests as reflecting American ideals of fairness and efficiency in educational opportunity).

181. Id. at 105. Historians note that “[b]etween 1820 and 1860 American cities grew at a faster rate than in any other period in U.S. history, as the number of cities with a population of over 5,000 increased from 23 to 145. That same period saw an average annual immigration of roughly 125,000 newcomers[.]” Id. The resulting expansion of the student population “increased the taxpayers’ burden and created new institutional demands for efficiency similar to those that governed the evolving nature of many American institutions.” Id. at 106. In response, schools sought to “demonstrate sound fiscal practice . . . by organizing themselves according to principles of bureaucratic management.” Id.

182. Id. at 110 (citation, internal quotation marks, and emphasis omitted).

183. Id.

184. Id. at 116. Critics have long emphasized the need for assessment methods that embrace complexity of thought and value expression beyond quantifiable metrics. See id. Lamenting the state of the education system in 1922, educational reformer John Dewey observed, “Our mechanical, industrialized civilization is concerned with averages, with percents. The mental habit which reflects this social scene subordinates education and social arrangements based on averaged gross inferiorities and superiorities.” Id. These concerns extend to standardized written exams, which critics argue disproportionately benefit students from privileged backgrounds with greater access to test preparation resources. Id. Further, an emphasis on outcomes over learning can heighten stress, encourage academic dishonesty, and create pressures that extend beyond students to parents, guardians, and even athletic coaches. See generally Joshua Lens, Operation Varsity Blues and the NCAA’s Special Admission Exception, 31 J. Leg. Aspects Sport 147 (2021). Following the Federal Bureau of Investigation’s “Operation Varsity Blues” investigation, several parents were accused of securing extra time, obtaining correct answers, or arranging for others to take exams on behalf of their children. Id. at 155. While oral exams can also induce stress and anxiety, they are less susceptible to manipulation and may better preserve the integrity of assessment. See Mariano et al., supra note 154, at 60 (“Oral examinations provide instructors with a method of assessment that all but eliminates academic cheating and plagiarism”).

185. See Off. Tech. Assessment, supra note 179, at 103. Education researchers Doug Archbald and Fred Newmann asked, “Why aren’t essay and oral exams used more frequently in high schools?” Archbald & Newmann, supra note 178, at 13. They attribute this scarcity to time constraints, noting that “students often have few opportunities to write or speak more than a sentence or two.” Id. Reflecting on the implications, they observe, “This is unfortunate, since most people would probably agree that the best way to determine whether a person understands a subject or problem is simply to ask him or her to explain and to respond to questions that the explanation itself is likely to provoke.” Id. (emphasis omitted).

186. See Naughton, supra note 178, at 290–91. In mathematics courses, oral exams are used to evaluate students’ understanding of both procedural and conceptual aspects of problem-solving. See Mariano et al., supra note 154, at 60–61 (citing Ralph Boedigheimer, Michelle Ghrist, Dale Peterson & Benjamin Kallemyn, Individual Oral Exams in Mathematics Courses: 10 Years of Experience at the Air Force Academy, 25 Primus 99 (2015)). In business education, they serve to assess both communication skills and content mastery. Id. (citing L.A. Burke-Smalley, Using Oral Exams to Assess Communication Skills in Business Courses, 77 Bus. & Prof. Comm. Q. 266 (2014)). In engineering, oral exams help measure students’ reasoning behind their problem-solving steps. Id. at 62 (citing Doug Belkin, As AI-Enabled Cheating Roils Colleges, Professors Turn to an Ancient Testing Method: Oral Examinations, Which Date at Least to Ancient Greece, Are Getting New Attention, Wall St. J. (June 1, 2023), https://www.wsj.com/us-news/education/ai-colleges-cheating-oral-exams-286e0091). At the doctoral level, oral defenses remain a standard requirement across most disciplines, with candidates presenting and defending their dissertations before a panel of experts. Jennifer Harrison, From Nerves to Triumph: Your Personal Guide to Dissertation Defense, The PhD Place (Aug. 26, 2023), https://thephdplace.com/guide-to-dissertation-defense [https://perma.cc/4HWX-6F48].

187. See generally John M. Burman, Oral Examinations as a Method of Evaluating Law Students, 51 J. Legal Educ. 130 (2001); see also András Jakab, Dilemmas of Legal Education: A Comparative Overview, 57 J. Legal Educ. 253 (2007).

188. Burman, supra note 187, at 132.

189. Id. at 134.

190. Id.

191. Id.

192. See id. at 138–40.

193. Id. at 135–39.

194. Id. at 138.

195. See Mariano et al., supra note 154, at 60–62 (discussing the rationale for oral exams).

196. See id.

197. See id. at 62.

198. See id. at 61.

199. See id. Additionally, oral exams offer pedagogical value because they are highly adaptive, allowing professors to “adjust their assessments based on their personal knowledge of the student and the student’s progression throughout the school year.” Naughton, supra note 178, at 291.

200. See Mariano et al., supra note 154, at 61. “Once students see their grade on a written exam, there is no guarantee that they will read all the comments[,]” whereas “[d]uring an oral exam, however, students will be present for the instructor’s immediate feedback.” Id.

201. See id. at 61. Oral assessments are described as vocationally relevant for “building confidence, enhancing communication skills, developing critical thinking skills, improving information-gathering skills, and fostering the ability to ‘think on their feet.’” Id.

202. See id. (noting that “[t]he oral examination provides the opportunity for insights into student thinking during the examination process”); see also Davis, supra note 5, at 10–17 (exploring questions about ethos and authenticity).

203. See Suzanne E. Rowe, Legal Research, Legal Writing, and Legal Analysis: Putting Law School into Practice, 29 Stetson L. Rev. 1193, 1202 (2000) (“Most legal writing courses teach objective and persuasive writing in different semesters”).

204. See id. at 1194, 1201 (noting that some programs also focus on client letter drafting and legislative skills).

205. Although the oral exam serves a different purpose from the practice oral argument, students may still benefit from brief opportunities to rehearse the oral exam format. Programs concerned about scheduling can offer short practice sessions alongside, or even in conjunction with, practice oral arguments.

206. See 14 C.F.R. § 61.43 (providing private pilot testing requirements); 14 C.F.R. § 135.293 (providing commercial pilot testing requirements); see also Aviation Instructor’s Handbook, supra note 109, at ch. 6, 6-1 (highlighting the importance of how assessments are structured and evaluated). The certified flight instructor (“CFI”) practical exam is widely regarded among aviators as one of the most demanding FAA check-rides. Jason Blair, Initial CFI Checkride Myths, Dispelled, AOPA (Apr. 12, 2018), https://www.aopa.org/training-and-safety/flight-schools/flight-school-business/newsletter/2018/april/12/initial-cfi-checkride-myths-dispelled. Unlike other practical tests, the CFI oral exam requires candidates not only to demonstrate mastery of aeronautical knowledge and flight technique but also to explain and teach those concepts effectively to student pilots. See id. This pedagogical component represents a qualitatively distinct and higher standard of competency. See id.; see also Aviation Instructor’s Handbook, supra note 109, at ch. 8 (discussing aviation instructor responsibilities and professionalism). In 2025, examiner-conducted original CFI practical tests had an approval rate of 73.7 percent, lower than the rates for any of the pilot certificate categories examined. Fed. Aviation Admin., supra note 148, tbl. 19 (reporting separate totals for flight instructor certificates and pilot certificates).

207. See Rowe, supra note 203, at 1202 (noting that the oral argument typically occurs in the second semester). Even before the advent of generative AI, some legal writing professors used short “Meetings with the Partner” simulations to verify that students had personally completed their research and drafting. Sourcebook 3d ed., supra note 25, at 206.

208. Aviation Instructor’s Handbook, supra note 109, at ch. 5, 5-16.

209. Id. at ch. 6, 6-11.

210. Id.

211. Id. at ch. 6, 6-11–6-12.

212. See id.

213. Duquesne Kline School of Law, Legal Research and Writing I Course Materials 2–4 (Fall 2025) (on file with author).

214. Id.

215. Id. at 4.

216. In the aviation context, an evaluator might ask a learner “why flying with an inappropriate mixture setting is bad.” Aviation Instructor’s Handbook, supra note 109, at ch. 3, 3-11. This question requires the student to explain that using the wrong fuel-air mixture can make the engine run poorly or cause it to be dangerously hot. Id. The Aviation Instructor’s Handbook provides a vivid illustration of check-ride readiness through the story of “Beverly,” a learner who begins training unable to answer questions that probe understanding, easily loses her place when interrupted, and becomes visibly frustrated by errors. See id. at ch. 3, 3-1. Months later, “she applies her knowledge to solve the problems” posed by her instructor, performs skills with confidence, and immediately detects and corrects small mistakes. Id.

217. Questions may be calibrated to greater sophistication depending on the problem vehicle. For instance, drawing on the Aviation Instructor’s Handbook’s principles for scenario-based oral evaluation, a legal writing check-ride might ask: “You located two cases that define ‘recklessness’ differently, one binding and one persuasive. Walk me through how you determined which case controlled your analysis, what additional research steps you considered, and how you would explain the resulting rule synthesis to a supervising attorney or judge.”

218. Aviation Instructor’s Handbook, supra note 109, at ch. 6, 6-12–6-13.

219. Id. Risk management is also a core component of pilot oral exams. The FAA requires candidates not only to demonstrate knowledge and skill but also to explain how they mitigate risks in realistic operational scenarios. See id. at ch. 1. Consider, for example, an instructor conducting an instrument check-ride who disables the autopilot on final approach at approximately 700 feet above the ground in rain and fog down to landing minimums. The candidate must demonstrate the ability to stabilize the aircraft and either complete a safe landing or execute a safe missed approach under considerable time pressure and environmental strain. Developing and maintaining the proficiency required to manage this scenario is no small endeavor, and it illustrates why aviation relies on check-rides as a countermeasure to automation-related risks. These dimensions raise broader questions about whether similar risk management principles, which are often taught implicitly, should be incorporated more intentionally into the law school curriculum.

220. Id. at ch. 3, 3-6.

221. Marie Summerlin Hamm, Benjamin V. Madison, III & Ryan P. Murnane, The Rubric Meets the Road in Law Schools: Program Assessment of Student Learning Outcomes As A Fundamental Way for Law Schools to Improve and Fulfill Their Respective Missions, 95 U. Det. Mercy L. Rev. 343, 375 (2018).

222. Id.

223. Id.

224. Id.

225. Id. at 376–77. It is important to note that “[s]ubstantive knowledge of law is but a starting point. The task of legal educators is to help students progress beyond rote learning to sophisticated analysis of multiple concepts, synthesizing principles, and having some idea of how to use their analytical skills in the multitude of tasks lawyers handle.” Id. at 376. Oral exam rubrics should reflect this pedagogical aim and, perhaps, are better suited than written assessments to evaluate such higher-order thinking skills. See Mariano et al., supra note 154, at 60 (noting that oral exams help build skills “useful in future job interviews, success in the workplace, and promoting the students’ ability to engage in civic discourse”).

226. See Clark & DeSanctis, supra note 30, at 4.

227. Id. at 7–9.

228. Id. at 19–20.

229. Id. at 13–15.

230. Borman, supra note 31, at 714–16.

231. Id. at 714. Borman also characterizes rubrics as “reductionist,” raising several concerns about their impact on student learning. Id. at 736–38. Among these are that “[s]tudents focus too much on the point categories instead of the ‘big picture,’” that rubrics encourage “students to approach assignments in a paint-by-numbers manner,” and that “[s]tudents can become too focused on the individual components rather than the overall scope and organization of the assignment.” Id. at 738 (citing Appendix 1, Survey Results). She also characterizes rubrics as “deterministic,” raising concerns that they assume monolithic categories and strengths or weaknesses for all students. Id. at 738–41.

232. See id. at 735 (“A checklist encourages one-dimensional, black-and-white thinking and does not develop critical reasoning skills”).

233. Id. at 735–36.

234. Id. at 745.

235. Id.

236. Id.

237. See Sourcebook 3d ed., supra note 25, at 206–07.

238. Id.

239. See Rachel Stabler, All Rise: Pursuing Equity in Oral Argument Evaluation, 101 Neb. L. Rev. 438, 465 (2023). A pass/fail system must require students to meet predetermined performance standards; otherwise, the exercise devolves into a grade-by-completion method that does not evaluate performance at all. Id.

240. Id.

241. See id. at 465–66.

242. Hamm et al., supra note 221, at 380; see also Stabler, supra note 239, at 465 (suggesting that professors could also recognize high-performing students through a “high pass” designation or an excellence award); Clark & DeSanctis, supra note 30, at 7 n.13 (describing a grading system including High Pass, Pass, Low Pass, and Fail).

243. See Aviation Instructor’s Handbook, supra note 109, at ch. 6, 6-5–6-7 (discussing rubric-based performance criteria for evaluating higher-order thinking skills, decision-making, and judgment).

244. Id.

245. Id. at ch. 6, 6-5.

246. Id.

247. Id.

248. Id.

249. Id. This gradation offers useful flexibility for instructors, as it avoids assigning a failing label to students who do not demonstrate certain learning standards, and accommodates the fluid nature of oral exams by allowing professors to refrain from grading areas not covered during the session. See id. at ch. 6, 6-5–6-6. In other words, it “is a form of learner-centered grading.” Id.

250. See id. at 6-6–6-7.

251. Id. at ch. 6, 6-7.

252. Id.

253. Id.

254. Id.

255. See id. The primary purpose of the oral exam is to verify that students can meet predetermined learning outcomes independently of AI assistance. Mariano et al., supra note 154, at 65. The act of self-assessment, however, may itself deepen learning and reinforce higher-order thinking skills. Aviation Instructor’s Handbook, supra note 109, at ch. 6, 6-7.

256. Aviation Instructor’s Handbook, supra note 109, at ch. 6, 6-7.

257. See Sourcebook 3d ed., supra note 25, at 67 (“In the LRW course, students learn to ‘think like a lawyer’”).

258. Under FAA regulations, a candidate who does not pass a practical test receives a notice of disapproval and must undergo additional training on the deficient tasks before retesting. 14 C.F.R. § 61.43, 61.49 (2026).

259. See Mariano et al., supra note 154, at 63–64 (discussing limitations and weakness of oral exams).

260. Id.

261. See id.

262. Id.

263. Id.

264. Stabler, supra note 239, at 448. Evaluating oral exams is difficult because it requires the examiner “to subjectively evaluate the performance of a student whose identity is known to the evaluator.” Id. This dynamic opens the door to “a variety of biases . . . , including cognitive biases unrelated to stereotypes, like the halo effect and conformation bias, as well as implicit biases arising from stereotypes associated with the student’s race, gender, or appearance.” Id.

265. See id. at 448–51.

266. Id. at 448–49. Notably, this phenomenon can arise even under anonymized conditions. Id. at 449. For example, when a law professor grades anonymous written exams sequentially, and the exam includes multiple questions, the professor may be inclined to score subsequent answers more favorably if the student’s first response was particularly strong. Id.

267. See Mariano et al., supra note 154, at 64 (“Standardized rubrics are important for oral exams given by multiple instructors/raters, which helps prevent subjectivity or bias in grading”).

268. Part III.D.3 devotes significant attention to understanding the use of rubrics in this context, and the Aviation Instructor’s Handbook also emphasizes their value in supporting authentic evaluation. See Aviation Instructor’s Handbook, supra note 109, at ch. 6, 6-6–6-7 (explaining that “a rubric is a guide for scoring performance assessments in a reliable, fair, and valid manner”). Another way to mitigate this concern is to involve, when possible, a panel of examiners who can offer diverse perspectives. See Mariano et al., supra note 154, at 63 (noting that “[h]aving assessors review oral exam questions . . . may reduce bias”).

269. See id. at 62–63 (discussing best practices for reducing the impact of anxiety on oral exam performance).

270. Id. at 63.

271. Id. at 62–63.

272. Id.

273. Id. Professors may decide to grade the oral exams on a pass/fail basis to reduce performance pressure. Id. at 63. For those interested, Ungrading: Why Rating Students Undermines Learning (and What to Do Instead) (Susan D. Blum ed., 2020) offers a compelling critique of the traditional grading system.

274. See Mariano et al., supra note 154, at 63 (explaining “that oral exams provide students with opportunities for managing modest amounts of stress, similar to that which they will face in their professional careers”).

275. Brian D. Zanger, Using Oral Exams in Physics and Astronomy Courses, at 4, https://doi.org/10.48550/arXiv.2509.09846 [https://arxiv.org/pdf/2509.09846] (preprint, Sep. 11, 2025).

276. Id.

277. Id.

278. Mariano et al., supra note 154, at 62.

279. See 42 U.S.C. §§ 12131–12165 (requiring public colleges and universities to provide reasonable accommodations to qualified students with disabilities); see also id. §§ 12181–12189 (imposing similar obligations on private institutions classified as places of public accommodation).

280. Suzanne E. Rowe, Reasonable Accommodations for Unreasonable Requests: The Americans with Disabilities Act in Legal Writing Courses, 12 Legal Writing 3, 51 (2006).

281. Id. at 52.

282. Id.

283. Id.

284. Id.

285. Id.

286. Id.

287. Id.

288. Recall that oral exams were largely abandoned due to logistical constraints, rather than pedagogical shortcomings. See supra notes 179–186 and accompanying text. This context makes it especially important to consider these strategies. See Mariano et al., supra note 154, at 64 (admitting that “effective implementation of oral examinations requires forethought, preparation, and hard work on the part of the instructor”).

289. Id. One model involved “cooperative oral final examinations,” in which “students worked in groups of four and one grade was given to the group for the exam.” Id. Students received “a list of possible questions” several days in advance and “prepare[d] for the exam together.” Id. These groups of students “watched a teaching video and were asked questions about theories and practices.” Id. For this cooperative exam exercise, “each group met with two instructors for 15 minutes.” Id.

290. Id. at 64–65 (discussing specific recommendations for online education).

291. Id. at 63.

292. See infra Part III.F (examining this potential new approach).

293. It is important to remember that “[o]ral examinations are not a ‘one size fits all’ solution to the challenges of artificial intelligence chatbots, but they also are not something to fear.” Mariano et al., supra note 154, at 65.

294. Moreover, “[t]he oral exam is simply an additional tool in the educator’s toolbox—and in certain situations, it is precisely the tool that instructors and students need.” Id.

295. Compare Louis J. Sirico, Jr., Teaching Oral Argument, 7 Persps.: Teaching Legal Rsch. & Writing 17, 17 (1998) (emphasizing the importance of oral argument as a means to teach students how to structure and deliver a persuasive legal argument), with Allison S. Theobold, Oral Exams: A More Meaningful Assessment of Students’ Understanding, 29 J. Stat. & Data Sci. Educ. 156, 156 (2021) (describing oral exams as “a powerful way to gauge student understanding . . . allowing for genuine conversations about students’ understandings”).

296. Ben Bratman, Legal Research and Writing as a Proxy: Using Traditional Assignments to Achieve a More Fundamental Form of Practice Readiness, 25 The Second Draft 7, 8 (2011) (noting that almost three-quarters of legal writing programs require an oral argument); see also Rowe, supra note 203, at 1215 n.45 (noting that “almost all students consider the experience a highlight of the first year of law school”).

297. Bratman, supra note 296, at 7.

298. Id.

299. Sirico, supra note 295, at 17.

300. Theobold, supra note 295, at 158.

301. See id.

302. See Enrollment Dataset, Analytix, https://analytix.accesslex.org/download-dataset (last visited July 7, 2026) (select a year in the “Enrollment” column, then click “Download Excel”). According to the AccessLex Institute, 196 ABA law schools reported enrolling 43,726 first-year students in 2025, yielding an average entering class size of approximately 223 students per school. See id.

303. See id. (providing enrollment data). These calculations assume an eight-hour workday and are based on the AccessLex Institute’s enrollment data.

304. Cuban Interview, supra note 15.

305. Id.

306. See id.

307. See Aviation Instructor’s Handbook, supra note 109, at ch. 3, 3-20–3-21 (discussing right-brain/left-brain science); see generally Muriel Deutsch Lezak, Daine B. Howieson, David W. Loring, H. Julia Hannay & Jill S. Fischer, Neuropsychological Assessment (4th ed. 2004); Jaak Panksepp, Affective Neuroscience: The Foundations of Human and Animal Emotions (1998); Allan H. Ropper & Robert H. Brown, Adams & Victor’s Principles of Neurology (8th ed. 2005).

308. Aviation Instructor’s Handbook, supra note 109, at ch. 3, 3-20–3-21.

309. Id.

310. Id.

311. Id. at ch. 3, 3-20.

312. Id.

313. See Paula Franzese, Law Teaching for the Conceptual Age, 44 Seton Hall L. Rev. 967, 970–71 (2014).

314. Id. at 971.

315. Id. (citation and internal quotation marks omitted).

316. Id.

317. Id.

318. Franzese contends that “we [are] in the midst of a paradigmatic shift away from the more linear ‘information age’ (where left-brain skills predominated) to the more innovative ‘conceptual age,’” and that statement seems truer than ever now. Id.

319. Id.

320. See 14 C.F.R. § 61.43(a).

321. For example, software engineering curricula teach “design thinking,” which “is an iterative, human-centered approach to problem solving” that law schools might adapt to help students refine legal arguments and develop client-focused solutions. Jacqueline E. McLaughlin, Elizabeth Chen, Danielle Lake, Wen Guo, Emily Rose Skywark, Aria Chernik & Tsailu Liu, Design Thinking Teaching and Learning in Higher Education: Experiences Across Four Universities, 17 PLOS One e0265902, at 1 (2022), https://doi.org/10.1371/journal.pone.0265902 [https://perma.cc/33HL-EK2G]; Rebecca Linke, Design Thinking, Explained, MIT Sloan (Sep. 14, 2017), https://mitsloan.mit.edu/ideas-made-to-matter/design-thinking-explained [https://perma.cc/Y8EU-2YF3].

322. Burman, supra note 187, at 134.

323. Naughton, supra note 178, at 291.

324. This raises broader questions about whether Bloom’s framework itself must adapt to contemporary realities shaped by generative AI, an issue that merits further consideration.