In a field undergoing rapid technological transformation, how can law schools address the ever-widening gap between “thinking like a lawyer” and actually being ready to practice? Clinical education is one of the few places where students are already asked to make that leap through representing real clients, making judgment calls, and owning the consequences of their work.

But the profession is changing. Major law firms are regularly announcing massive investments in generative AI infrastructure, and access-to-justice advocates are pushing for equally ambitious innovation. The repetitive, high-volume research, review, and drafting tasks that have long been fertile ground for training are being absorbed by AI. The runway from junior associate to midlevel is compressing, and traditional attorney mentorship frameworks are strained. Recent graduates will be expected to add value from the start of their careers, even as the old apprenticeship model is disappearing.

Law school clinics are one of the clearest models for teaching those skills. They forgo the billable efficiency that drives modern practice. They build on the doctrinal grounding students develop in the classroom and ask them to put that learning to work in messy, real-world contexts. They create space for students to build the foundations of good lawyering and to experiment with the humanity that live client work requires. That space is sometimes called “lawyering in slow motion,” in which students break the complex process of lawyering into discrete, actionable steps and explore the reasoning behind their choices. Here, they learn to work within the uncertainty of both the law and human relationships. And they begin to take internal ownership of those relationships and their own accountability as lawyers.

This makes clinics an appealing testing ground for working with AI tools in real practice settings, under the supervision of a qualified attorney-professor. But this instinct requires rigorous scrutiny. Clinic clients often have no safety net, and the consequences of a misstep may be far more personally significant than an error in billable practice. These are families fighting eviction, individuals seeking asylum, defendants engaging with the criminal justice system, and small business owners staking everything on their futures. The consequences can also carry over to the supervising attorney, who may suffer when a student – or an AI output – goes astray. Generative AI adds a new layer of risk to an already serious level of responsibility.

However, clinicians can lead and shape how AI is evaluated and, in some settings, integrated into the work of becoming practice-ready, because they are immersed in the world between teaching and consequence. Clinical students are often surprised by the uncharted gap between the legal rules learned in class and the judgment calls made alongside their clients. But this is where clinicians have always excelled – supporting students through the ambiguity and stepping back to let them learn when to trust the rules, when to question them, and how to explain those options to their clients. For clinics that opt to explore AI in their work, these frameworks can be the starting point for responsible, client-centered use. For others, the clinic can be where they work through and decide that certain tools may not be compatible with a particular matter or client community.

Unlike passive AI use, which risks eroding the thinking that makes a good lawyer, the clinical model demands that students plan before they act and take ownership of their work product, whether or not AI is used. These foundational skills are what employers say they need on day one: lawyers who can exercise judgment, solve problems creatively, and decide when, how, and which technology to use and supervise it effectively.

In the clinical setting, instead of accepting AI-generated content in the interests of saving time, students must do the work upfront to design, prompt, and interrogate it, building detailed verification processes to check for hallucinations, false completeness and omission, and contextual applicability and then stand up and defend the reasoning behind the output. Clinical work also gives students concrete experience with clients who may be using AI tools themselves (and fact-checking their attorneys’ legal advice in real time).

The opportunity goes beyond just technical preparedness. New lawyers are entering a profession and a world being transformed by AI. In clinics, they can grapple with what that means, raising questions about AI’s costs and biases and engaging with its impact on the communities clinics serve. For some students, that will solidify a calling toward public interest work. And clinics that opt not to use AI can also be a logical place to surface questions about technology, power, professional responsibility, and the lawyer’s role in shaping it.

The deans of many top law schools have said it repeatedly: good lawyering is ultimately defined by our humanity – the judgment, accountability, collaboration, and empathy inherent in a trusted advisor. As the basic knowledge work of lawyering is increasingly delegated to AI, those human capacities remain the core of lawyering itself. The clinical model already prioritizes these skills.

Nowhere else in legal education do students experience the real risks their clients face, shape strategy that affects real lives, or celebrate and commiserate alongside real people the way they do in clinics. Clinics are where abstract legal education comes to life. In an AI-infused profession, no setting prepares the next generation of lawyers more fully to practice and to understand the world they are practicing in.