In March 2026, Chief Justice John Roberts issued a warning that landed like a thunderclap in the legal profession. Speaking at a judicial conference, Roberts told young lawyers that artificial intelligence would fundamentally reshape their careers — not in some distant future, but now. The remarks, first reported by Business Insider, carried unusual weight. This wasn’t a Silicon Valley evangelist or a venture capitalist talking his book. This was the most senior jurist in the United States, a man who writes his opinions in longhand, telling the next generation of attorneys that the ground beneath them was shifting.
Roberts didn’t mince words. He acknowledged that AI tools were already performing tasks that once consumed the first several years of a young associate’s career — legal research, document review, contract analysis, due diligence. The implication was stark: the traditional apprenticeship model, in which junior lawyers learned their craft by grinding through thousands of billable hours on precisely these tasks, was eroding. And fast.
What made Roberts’s comments so striking wasn’t just the message. It was the messenger. The Chief Justice has historically been cautious about technology in the courts. In his 2023 year-end report on the federal judiciary, Roberts offered a measured assessment of AI, noting both its promise and its risks, particularly the well-documented problem of AI “hallucinations” — instances where large language models fabricate case citations or invent legal precedents wholesale. By 2026, his tone had shifted from caution to urgency.
That shift mirrors what’s happening across the profession at startling speed.
Major law firms have spent the last two years integrating AI tools into their workflows with an intensity that would have seemed improbable even in 2024. Allen & Overy’s partnership with Harvey AI, announced in early 2023, was treated as a novelty at the time. Now it looks like the opening salvo. Firms including Latham & Watkins, Davis Polk, and Kirkland & Ellis have all deployed generative AI platforms for research, drafting, and case analysis. According to a 2025 survey by the American Bar Association, more than 35% of law firms with over 100 attorneys reported using some form of generative AI in daily practice, up from roughly 10% the year prior.
The economics are impossible to ignore. A first-year associate at a top firm bills at $400 to $600 per hour. An AI tool can perform certain research and review tasks in minutes that would take that associate days. For clients who have long complained about the inefficiency of the billable hour model — and for firms under pressure to demonstrate value — the calculus is straightforward. The question isn’t whether AI will replace certain junior lawyer functions. It already has.
But Roberts’s warning contained a subtlety that many initial reactions missed. He wasn’t simply predicting job losses. He was describing a transformation in what it means to be a competent lawyer. The skills that matter are changing. Technical legal knowledge remains essential, but so does the ability to work with AI systems — to understand their outputs, spot their errors, and direct them effectively. Roberts suggested that lawyers who can’t do this will be at a severe disadvantage.
This is where the profession’s anxiety is most acute.
Law schools are scrambling to respond. Stanford Law School, Georgetown, and Harvard have all introduced courses on AI and legal practice within the last 18 months. Suffolk University Law School in Boston launched a certificate program in legal technology. But curriculum changes take time, and the technology isn’t waiting. Several legal educators have privately expressed concern that the pace of AI development is outstripping the academy’s ability to prepare students for what they’ll encounter in practice.
There’s also the access-to-justice dimension that Roberts touched on, and that deserves more attention than it typically gets. The United States has an enormous unmet need for legal services. Roughly 80% of civil legal needs among low-income Americans go unaddressed, according to the Legal Services Corporation. AI-powered tools — chatbots that can help people fill out court forms, platforms that can draft simple wills or landlord-tenant complaints — could begin to close that gap. Some already are. The nonprofit organization Free Law Project and platforms like DoNotPay have been expanding their AI-driven services for people who can’t afford attorneys. Roberts acknowledged this potential, framing it as one of AI’s most significant contributions to the justice system.
Not everyone is convinced the optimism is warranted. Critics point to the hallucination problem, which remains stubbornly persistent despite improvements in model accuracy. In a widely cited incident from 2023, a New York attorney submitted a brief containing fabricated case citations generated by ChatGPT. The attorney was sanctioned. Similar incidents, though less dramatic, continue to surface. Judges in multiple federal districts have implemented standing orders requiring lawyers to disclose the use of AI in filings and to verify the accuracy of any AI-generated content. The Southern District of Texas, the Northern District of Illinois, and the Eastern District of Pennsylvania all have such orders in place.
So the profession finds itself in a peculiar position. AI is simultaneously the most promising tool for expanding legal access and the most immediate threat to the traditional career path of young lawyers. Both things are true at once.
The big firms are adapting by redefining what junior associates do. Instead of spending their first years reviewing documents in windowless conference rooms, new associates at some firms are being trained as AI supervisors — reviewing and refining the output of machine learning tools, focusing on higher-order analytical work earlier in their careers. Proponents argue this is actually better for professional development. Skeptics counter that something is lost when young lawyers skip the tedious groundwork that builds deep familiarity with case law and legal reasoning.
There’s real tension there, and Roberts seemed aware of it.
His comments also arrived against a backdrop of broader anxiety about AI’s impact on white-collar professions. A March 2025 report from Goldman Sachs estimated that generative AI could affect roughly 300 million full-time jobs globally, with legal services among the sectors most exposed. McKinsey’s research has projected that up to 23% of lawyer work hours could be automated with current technology. These are not speculative figures from fringe forecasters. They’re mainstream estimates from institutions the legal and financial worlds take seriously.
The regulatory picture adds another layer of complexity. The European Union’s AI Act, which began phased implementation in 2025, classifies certain AI applications in the justice system as “high risk,” subjecting them to stringent transparency and accuracy requirements. In the United States, regulation has been more fragmented. The Biden administration’s 2023 executive order on AI safety laid some groundwork, but comprehensive federal legislation has not materialized. State bar associations have been left to craft their own guidance, leading to a patchwork of rules that varies significantly by jurisdiction.
California’s state bar issued ethics guidance in late 2025 that essentially told lawyers: you can use AI, but you’re responsible for everything it produces. Full stop. That principle — human accountability for machine output — has become the consensus position among bar associations nationwide. It sounds simple. In practice, it means lawyers need to understand AI well enough to catch its mistakes, which circles back to Roberts’s core point about competency.
The market for legal AI technology is booming. Thomson Reuters acquired Casetext and its AI assistant CoCounsel in 2023 for $650 million. LexisNexis rolled out its own generative AI tools. Startups like Harvey, EvenUp, and Ironclad have raised hundreds of millions in venture capital. The legal tech sector attracted approximately $1.2 billion in funding in 2025, according to data tracked by Crunchbase — a record. Money is pouring in because the opportunity is enormous. The global legal services market exceeds $900 billion annually, and much of it runs on processes that haven’t fundamentally changed in decades.
And yet, for all the investment and hype, the technology’s limitations are real. Large language models don’t understand law. They predict text. They’re extraordinarily good at pattern matching and generation, but they lack the judgment, ethical reasoning, and contextual understanding that define good lawyering. They can draft a contract clause, but they can’t tell you whether the deal makes strategic sense. They can find relevant precedent, but they can’t craft the narrative that wins at trial.
This is where Roberts’s message carries its deepest resonance. He wasn’t telling young lawyers to fear AI. He was telling them to be better than AI. To develop the judgment, creativity, and human skills that machines can’t replicate. The lawyers who thrive, he suggested, will be those who treat AI as a powerful tool and not a replacement for thinking.
That’s easier said than done when you’re a 26-year-old with $200,000 in law school debt watching a chatbot do in seconds what you spent three years learning to do.
The legal profession has weathered technological disruption before. Westlaw and LexisNexis transformed legal research in the 1970s and ’80s. Electronic discovery upended litigation practice in the 2000s. Each time, the profession adapted — sometimes slowly, sometimes painfully, but it adapted. The difference now is speed. Previous waves of legal technology took years or decades to reshape practice. Generative AI is doing it in months.
Roberts, who turns 72 in January 2027, has been Chief Justice since 2005. He’s seen the Court grapple with technology cases from cellphone searches to social media regulation. His willingness to speak so directly about AI’s impact on the profession signals that the judiciary itself recognizes the magnitude of what’s happening. It’s one thing for tech companies to make grand claims about AI transforming law. It’s another when the Chief Justice of the United States says roughly the same thing.
The question now is what the profession does with the warning. Law firms, law schools, bar associations, and courts are all making decisions right now that will shape legal practice for decades. Some of those decisions will prove wise. Others won’t. But the one option that isn’t available — as Roberts made clear — is pretending nothing has changed.
Something has. And the lawyers who recognize that earliest will be the ones who define what comes next.
WebProNews is an iEntry Publication