AI training moves to law schools, at the cost of the associate apprenticeship

Legora is expanding its legal AI training programme to UK law schools and has signed up its first university partners, Non-Billable reports. The move puts education, rather than software deployment alone, at the centre of the AI-native law-firm question: who develops the lawyers capable of directing, checking and owning automated work?

That question matters because the advantage belongs to firms that grow lawyers faster, not simply to firms that hire fewer beginners. Training law students before they enter practice is at least consistent with that model. It treats AI capability as something lawyers must acquire early, rather than as a substitute for the judgment that legal work eventually requires. The cost is that an associate apprenticeship cannot remain an unexamined byproduct of routine work. If junior tasks are automated without a deliberate way to build judgment, a leaner firm may also be a firm with fewer people ready to become senior lawyers.

The strongest objection is that firms may simply stop training associates. Jordan Furlong’s account says the answer to how firms will train associates after AI is that they will train partners instead. Legora’s programme does not resolve that objection. It establishes that a legal technology company is extending AI training into UK law schools; it does not establish that the participating schools, firms or students will develop stronger legal judgment, nor that the programme leads to better lawyers. Nor does it show that associate work has been removed from firms. The stronger reading would require observable outcomes: graduates using AI responsibly on legal matters, firms assigning them progressively harder work, and later evidence that they become capable senior lawyers. None of that is reported here.

A separate LawNext podcast report points in the same direction without proving the result. It says Alex Su was named president of Latitude Legal and that Latitude reported a 52 percent jump in new law firm and corporate clients in the first half of the year. The report also describes Su’s view that better AI would increase demand for lawyers with real judgment to direct it, check it and own the work. That is a company-reported number and an attributed prediction, not a demonstration that the legal sector is developing lawyers faster. It does, however, identify the measurable demand that a training strategy must meet: not more people who can perform routine tasks, but more people who can take responsibility for complex work.

The next confirming fact is not another training announcement. It is a published outcome from Legora’s university partners showing what students learned, how they used the system, and whether those students moved into progressively more judgment-intensive legal work. A later hiring or promotion figure tying that training to capable lawyers would support the stronger reading. If the programme produces familiarity with a tool but no reported change in legal responsibility, the announcement will have shown distribution of AI education, not faster development of lawyers. We will count the result, not the launch.