WASHINGTON, DC — One general counsel at the recent Thomson Reuters Institute’s 25th Annual Law Firm COO & CFO Forum told the room about a recent dinner with a table of first-year law students. He described them as terrified: about to take on enormous debt for a degree in a profession they suspect AI will hollow out.

He wasn’t the only speaker to carry that fear into the room. One panelist described a first-year class being asked to use AI for tasks they don’t yet understand, and admitted that now she was nervous, too. An attendee during a separate Q&A spoke about his own son’s fears of finding a career in an AI-infused world.

Then came the reassurance. Over two days, law firm leaders lined up to say the same thing. First- and second-year associates aren’t going away. They can’t be replaced; and, in fact, they’re the future of the profession. One leader looked ahead to 2031 and saw a high-value proposition for large firms, provided they make space for new talent.

These firm leaders meant it, no doubt; but yet, looking at what those leaders in the room actually did, it’s not hard to see why those law students are scared.

What the budget says

On Day 1, attendees played a game. Each table got a fixed budget, a menu of initiatives, and successive rounds of shocks that forced cuts. The facilitator’s premise was that law firms have long survived by betting next year will look enough like last year to defer the hard decisions. This time, no deferring.

The results were remarkably consistent. Nearly every table protected the AI-assisted document review and drafting platform. Associate compensation realignment was swapped out, trimmed, or never funded to begin with. When the facilitator asked what every team had protected, associate compensation didn’t make the list.

One attendee explained why: They were tired of fighting the associate compensation fight, having done it for years.

The legal industry facing some simple, if not altogether comfortable truths: Law firms aren’t eliminating associates; but they are letting AI dismantle the apparatus that trained junior associates, and only a handful of firms have started building a new apparatus to take its place.

It was a game, but the choices echoed what panelists described in real life. One CFO noted that clients now refuse to pay for summer associates’ hours and openly wondered whether first- and second-years are next. A COO said that when someone leaves, the firm now thinks hard about what that role should look like going forward — a polite way of saying not every departure gets replaced. Another leader acknowledged another unfortunate reality, that the organizational model and hiring are going to change.

Law firm clients aren’t softening the message, either. The general counsel who had dinner with those law students said his legal department now uses AI to generate first drafts, such as motions to dismiss, then hands them to outside counsel, telling the firm it doesn’t need a first-year associate to do the drafting or research. Other panelists described clients having moved quickly from a mantra of Don’t use AI to expecting it, without wanting to pay for it.

Nobody at the Forum said they planned to stop hiring associates, of course; and one law firm leader said the firm’s summer and first-year classes are actually getting larger. However, when the room had to decide where a hypothetical law firm’s money went, associates kept losing.

Dismantling in the apparatus

The bigger threat, however, isn’t that law firms will eliminate associates. It’s quieter than that.

For generations, the work for which clients didn’t want to pay full price — such as document review, first drafts, and long hours of research — was also how junior lawyers learned. The billable grind wasn’t just a business model, it was the training program. AI is absorbing exactly that work, faster than firms can decide what will replace it. The GC who met those 1Ls recalled spending three months of his first year as a junior lawyer in warehouses, going through boxes of documents for a jury trial. That work, he said, is probably all gone now.

One CFO laid out the problem with refreshing candor. Firms need to be strategic about how associates learn now. Should they learn to draft without AI first? Only with it? Some hybrid? She admitted she’d asked a lot of questions without answers, which drew some nervous laughs, because that is precisely where the legal industry now stands.

Another firm leader’s answer was getting associates back in the office to learn from senior associates and partners because proximity matters. However, sitting near a partner teaches far less when the hours of work that used to bring you into that partner’s office now take a machine just minutes to complete. And a first-year class prompting AI through tasks they don’t yet understand isn’t training either.

Which leaves the industry facing some simple, if not altogether comfortable truths: Law firms aren’t eliminating associates; but they are letting AI dismantle the apparatus that trained junior associates, and only a handful of firms have started building a new apparatus to take its place.

Building the simulator

That handful does exist, however — and fortunately for the legal industry, it’s growing.

At Legalweek earlier this year, AI-driven training simulations were an early, promising idea. Such simulations could provide deposition practice with real-time feedback, and a chance to fail behind closed doors instead of in front of partners. By this Forum, just seven month later, one firm’s innovation leader was describing them as a deliberate model.

We don’t let junior pilots fly planes without plenty of simulator time, she argued, and lawyers should be no different. Compress the repetitions, and build judgment on purpose instead of hoping it accumulates through grunt work.

That approach treats training as a cost of doing business, the way airlines treat pilots’ simulator hours, rather than something billable work has to subsidize. In an important development compared to where things stood at Legalweek earlier this year, it also comes with a demand thesis to fund it. AI is generating new legal work in litigation, copyright, regulatory, and employment matters, and that surge in demand hours should be put towards funding the new model.

One COO said his firm has spent four months tagging time entries by AI use and found roughly a 5% efficiency gain. He also found that clients mostly want firms to take on new kinds of work rather than to send smaller bills.

Another CFO framed the opportunity more simply: If AI saves an attorney 30 minutes a day, what happens to that time? Is it spent with the client? Or, it is spent mentoring? One option that was conspicuously not floated was just filling those hours with more work. For the legal industry, the omission is mildly revolutionary.

Two kinds of firms

Despite its promise, the simulator view was a minority one at the Forum. The more prominent posture was cost control, such as dashboards tracking token consumption, questions about whether a cheaper tool could handle a task, and even whether core systems such as document management are still needed at all. Those aren’t unreasonable instincts in an expensive technology cycle, of course; but they treat AI as a line item to minimize, not an industry-reshaping event that should spur a re-examination of how lawyers are made.

For law firms, the question of where the next generation of senior lawyers comes from may soon split the market, not by size or rates but by talent. A top graduate in 2027 will be choosing among law firms that treat her early years as a cost to be compressed and endured, and firms that treat those years as an investment to be accelerated.

So, the best advice for those terrified 1Ls might be simply, Don’t ask a firm whether it uses AI. They all do now. You should ask whether they’re using it to make you a better lawyer or merely a less expensive one.

You can register now for next year’s 26th Annual Law Firm COO & CFO Forum on Oct. 20-21, 2027, in New York City here‍