Generative AI vs law: Profession at crossroads

Generative artificial intelligence (AI) has been hailed as a potential game-changer for the legal profession, promising efficiency gains and broader access to justice. Yet, as a recent Bloomberg opinion piece highlights, the technology’s impact on lawyers and law firms may be far more complex — and disruptive — than initially imagined. While large-language models (LLMs) such as ChatGPT mimic many of the functions lawyers perform, they also raise thorny questions about reliability, economics, and the very future of professional services.

At the heart of the debate is the tension between efficiency and accuracy. Sean Fitzpatrick, chief executive officer of LexisNexis North America, UK, and Ireland, frames the issue succinctly: “As a matter of fact, they don’t really deal in the truth. They deal in probabilities, and then they make plausible arguments.”

This probabilistic reasoning can make AI-generated legal arguments sound convincing, even when they are riddled with fabricated case law or citations — a problem that has already embarrassed both attorneys and judges.

Despite these shortcomings, companies like LexisNexis and Thomson Reuters see an opportunity to harness AI responsibly. Both firms are developing legal AI platforms, trained on curated data to reduce hallucinations. Early research from Stanford scholars suggests these tools are indeed more accurate than ChatGPT, though still prone to error. Market signals indicate demand is strong: LexisNexis reported underlying revenue growth of 9% in the first half of 2025, while Thomson Reuters’ legal division recorded 8% organic growth in the same period.

Still, law firms themselves remain slow adopters. A recent Thomson Reuters survey found the legal sector lagging behind corporations in AI investment. This measured pace of adoption may reassure traditionalists who envision a “reassuringly familiar” future — a scenario described a decade ago by Richard and Daniel Susskind in The Future of the Professions. In this world, lawyers retain their traditional roles while standardizing and systematizing routine tasks with AI’s help.

But the Susskinds also outlined a second, more disruptive possibility: professional work could be steadily dismantled as machines displace human experts.

Daniel Susskind has since expanded on this theme in World Without Work, arguing that automation could fundamentally transform — and shrink — professions. Which path the legal sector takes may hinge on whether AI can ever be reliable enough to replace lawyers outright.

Even partial automation, however, threatens the unique economics of the legal profession. Law firms still largely rely on billing by the hour — a system that rewards time spent, not output delivered.

As Bloomberg notes, everywhere else, it means how much workers get done in a given amount of time. Despite decades of technological innovation, productivity in legal services, as measured by the U.S. Bureau of Labor Statistics, has actually declined since the 1980s.

Ironically, the first wave of AI adoption seems to have reinforced the billable-hour model. 

“We’re seeing high single-digit increases in rates per hour. I think part of that is attorneys are now delivering more value to their clients and they’re doing it in less time," Fitzpatrick observes.

Clients appear willing to pay a premium for higher-quality results, even if the work takes fewer hours.

Yet some analysts believe this trend is unsustainable. Bruce MacEwen and Janet Stanton of the consultancy Adam Smith, Esq. argue that reduced billable hours will eventually “kneecap the traditional hourly billing model.”

They foresee fixed-fee pricing becoming the norm, though they caution that most firms lack the financial infrastructure to price services effectively. The likely outcome, MacEwen predicts, is “a vast winnowing of traditional law firms.”

This raises a larger societal question: should lawyers be replaced at all? As the Susskinds point out, much of the world cannot afford first-rate professional services. If reliable AI could answer legal queries at scale, access to justice would improve dramatically.

But Jonathan H. Choi of Washington University School of Law warns that this vision remains distant. His research shows that LLMs excel at simple tasks but falter on complex, ambiguous questions, with results varying wildly depending on how prompts are worded. In his words, “Large Language Models Are Unreliable Judges.”

For now, then, lawyers appear safe from wholesale replacement, but not from disruption. Generative AI is already reshaping the economics of law, forcing firms to reconsider their billing practices and clients to rethink what legal expertise is worth. 

The profession’s future may lie somewhere between the Susskinds’ “reassuringly familiar” continuity and their prediction of dismantlement. Either way, as Bloomberg’s analysis suggests, the era of AI-driven law is no longer theoretical. It is here — and the legal profession must adapt.

By Sabina Mammadli