A 1L said it better than most senior partners have: “The profession seems more concerned with controlling AI than preparing students to use it well.”
The legal profession has spent years debating what to do about AI. The students who will practice law have already moved on.
The Thomson Reuters Institute’s 2026 Law Student Pulse Survey of 1,874 US law students might have expected to find a generation of reckless AI adopters – instead, it found an institution problem. More than half (57%) of students use AI several times a week or more for academic work. Among postgraduate students, that number reaches 84%. The pleasant surprise: students did not wait for guidance on AI. They self-taught, built their own frameworks, and arrived at a more nuanced position than most of their schools.
Consider what nuanced actually looks like in the survey. Three in four students say AI literacy counts as an essential professional skill. The same proportion say over-reliance on AI could undermine the critical legal skills they need. Holding both views at once without collapsing into either alarmism or blind enthusiasm, is precisely the kind of professional judgment law school is meant to cultivate. Students got there anyway. Without help.
The cost of institutional silence
Nearly a third say their school does not give them the AI skills they need for their careers. Among third year law students (3Ls)––those with the most real-world exposure through internships and clinics––only 30% feel adequately prepared. The more time students spend watching how the profession actually operates, the less confidence they have in what they learned.
Almost half of students report that AI policies vary by professor, leaving no consistent signal about what counts as permissible use, responsible use, or misconduct. One student described peers sharing AI study prompts in private, qualifying every mention with assurances they would never use it for graded work, even while using it for everything else. As one first year law student (1L) put it: “The fear of being accused of cheating has pushed AI usage underground, which is the worst possible outcome for everyone. Underground usage means no quality control, no shared best practices, no institutional guidance, and no ethical framework.”
That student identified the real cost of institutional silence more clearly than most administrators.
The talent pipeline problem and the AI factor
Students rate AI’s impact on independent judgment, learning through experience, and mentorship quality as net negatives. One student framed the compounding risk plainly: “I don’t think AI will ever replace senior lawyers, but it will render many junior lawyers redundant, which may in turn reduce the number of future senior lawyers.”
The survey data backs that concern. Nearly half of all students expect entry-level roles to decline as AI absorbs work that historically taught junior lawyers the basic mechanics of practice. The assignments that build foundational skills, the research, the drafting, the document review, flow to AI tools instead. The developmental pathway through which the profession reproduces its expertise starts to narrow.
The one exception in how students rate AI’s impact: new AI proficiency qualifications, which receive a net positive rating. They see the credential forming. They want to hold it. And they plan to choose employers accordingly. Among heavy AI users, more than half say the absence of professional-grade AI tools would factor into whether they accept a role.
What students are actually asking for: A framework
The survey’s open-ended responses point consistently toward the same request. Students want a clear line between when AI use is permitted, encouraged, and prohibited. Not a ban. Not a blanket endorsement. A framework, applied consistently, across coursework, clinic work, internships, and eventual practice.
They also want the profession to move past hallucination risk as its default framing for why AI is dangerous. Knowing AI can hallucinate is not the insight anymore. Every student in the survey already knows that. What they want is practical guidance on how to guard against it. One 1L said: “The conversation should not be ‘AI can hallucinate, so don’t use it,’ but rather ‘AI can hallucinate, so here’s what you need to do to guard against it.'”
The more advanced students go further. Those with genuine AI fluency have already identified a distinction the broader profession has not caught up to: the difference between a general-purpose tool and one built to professional standards. One postgraduate student wrote: “The legal profession is missing the distinction between general-purpose AI and fiduciary-grade AI. AI in law shouldn’t just be a productivity tool; it must be a professional-grade partner built on verified legal datasets.”
That observation did not come from a partner or a professor. It came from a student.
Next steps for the profession: AI frameworks and competencies
Law schools need institution-wide AI frameworks to replace the current patchwork of individual faculty policies, ones that tell students what AI use is appropriate in which context, distinguish between coursework and clinic work and legal practice, and measure whether students develop AI fluency alongside the foundational skills AI cannot replace.
Bar associations and accreditors need to define AI competency with the same clarity they bring to any other professional obligation.
Legal employers have the most immediate leverage and currently use the least of it. Firms that publish AI competency expectations for incoming associates give law schools a concrete target to build toward. Staying quiet just shifts the problem downstream.
The lawyers arriving now know what AI can do and where it breaks down. What nobody has given them is a map for using it with the judgment the profession demands.
The students who will define the future of legal practice are sophisticated, anxious, and waiting for the profession to meet them where they live.
A 1L said it better than most senior partners have: “The profession seems more concerned with controlling AI than preparing students to use it well.”
Join the conversation!