Artificial intelligence has fundamentally changed how legal work is produced. Work that once required hours of legal analysis can now be produced in minutes. That is an extraordinary gain in efficiency, but it also changes where the real value lies. The challenge is no longer producing an answer. It is knowing whether the answer is accurate, complete, and can be trusted.

The profession is beginning to respond to that challenge. In June 2026, Rhode Island joined a growing number of jurisdictions issuing guidance on lawyers’ use of artificial intelligence, reinforcing a principle that has always existed: technology can assist the work, but it cannot assume responsibility for it. Lawyers remain accountable for the accuracy of AI-assisted work and for exercising appropriate professional judgment. That guidance reflects a broader shift taking place across the profession. As AI lowers the cost of producing legal work, the competitive advantage increasingly shifts to those who can review, supervise, and exercise sound judgment.

That shift has implications well beyond professional responsibility. It is beginning to reshape the economics and talent model of law firms. For decades, law firms were built around a simple reality. High-quality legal work required skilled human effort. If a client needed a research memorandum, a contract review, or a brief, someone had to spend the hours producing it. Firms hired and trained large classes of associates because legal work was inherently labor-intensive.

Artificial intelligence changes that equation. Today, a well-crafted prompt can generate a first draft in seconds. Research can be synthesized almost instantly. Contracts can be reviewed in a fraction of the time they once required. The bottleneck has moved. It is no longer creating legal work. It is determining whether that work is accurate.

When Wrong Looks Right

Lawyers make mistakes and junior associates miss issues. What makes AI different is not that it gets things wrong. It is how convincing those mistakes can look.

A bad first-year associate draft usually contains clues: the analysis may be incomplete, the writing may be awkward, or citations may be thin. Something often signals that the work needs attention. AI-generated work frequently removes those signals. The language is polished and the structure is clean. Everything appears finished, which creates a different problem for reviewers. They are no longer reading rough drafts that invite scrutiny. They are reading work that appears complete even when it contains fabricated citations, faulty reasoning, or unsupported conclusions. 

That matters because errors can travel further before anyone catches them. A flawed analysis can be reused, cited, or incorporated into other work long before someone realizes there is a problem. As a result, the burden shifts upward. Senior lawyers should be spending less time improving work and more time determining whether they can trust it.

What Happens Inside Firms

The traditional pyramid was designed around work production. Junior lawyers performed large amounts of work. Midlevel lawyers refined it. Partners supervised it. Increasingly, firms need to become review organizations: AI generates an initial draft; a junior lawyer checks sources, citations, and basic reasoning; a midlevel lawyer pushes on assumptions and looks for weaknesses; and a partner decides whether the work is strong enough to put in front of a client or in front of a judge. The work moves through layers of review.

The limiting factor is how quickly someone with judgment can determine whether the draft should be trusted. 

Who Should Firms Hire?

Many firms still recruit using signals that made sense in a different environment: prestigious schools, strong credentials, polished communication, and confidence under pressure. Those qualities still matter. They are simply no longer enough.

The lawyers who become more valuable in this environment are the ones who keep asking questions after everyone else thinks the answer has been found. They check the citation. They test the assumption. They follow the logic one step further. They are harder to impress with polished work because they care more about whether it survives scrutiny. Some of the traits firms have traditionally rewarded can become liabilities if left unchecked. Confidence can turn into overconfidence. Speed can reduce examination. Fluency can create the appearance of understanding where little exists. The profession has spent decades identifying people who can produce answers. It now needs to become better at identifying people who can challenge them.

How Do Lawyers Learn Judgment Now?

This presents another obstacle. For generations, junior lawyers developed judgment through repetition. They reviewed documents, drafted motions, researched issues, and received feedback from more experienced lawyers. The process was often inefficient, but it worked.

If that work gets absorbed by AI, the next question becomes obvious: How does a lawyer learn judgment if fewer opportunities exist to exercise it?

The answer is unlikely to be more policy manuals or additional AI training sessions. Firms will need to create experiences that force lawyers to evaluate, challenge, and defend conclusions. That may involve reviewing intentionally flawed analyses. It may involve exercises built around finding errors rather than producing drafts. It may involve more direct coaching from senior lawyers than many firms currently provide.

Whatever the solution, firms cannot assume judgment develops automatically. The apprenticeship model depended on exposure to work. If the work changes, the training model must change with it.

Two Different Futures

Law firms appear to be moving in two different directions.

Some will use AI to deliver work faster and at a lower cost. Others will focus on something different: trust. Their pitch to clients will not be that they can generate an answer quickly. Everyone will be able to do that. Their pitch will be that their answer has been challenged, examined, and defended before it reaches the client. Both approaches can work. The mistake is assuming they require the same people, the same incentives, or the same management systems. They do not.

The Real Scarcity

The legal industry does not have a shortage of intelligent people. It never has. What it has now is a growing shortage of people willing and able to look at a polished answer and ask, “How do we know this is right?”

For decades, firms built their economics around producing legal work. Increasingly, their value will come from validating it. Artificial intelligence can draft a brief. It can summarize a case. It can produce an answer that sounds convincing. But, it cannot accept responsibility for that answer. A lawyer still has to do that. The firms that understand the difference will have a significant advantage over those that do not.