John Bliss and Johanna Schandera, 1/26/25
A recent survey by the National Law Review asked 67 experts in legal AI for their 2025 predictions. While the journal published the full responses, it offered no analysis or summary. This post is an effort to distill some general patterns in these forecasts. We focus on the roughly equal divide between those who expect radical transformation—a strange new world of AI co-workers—and those who foresee a year of gradual implementation.
Most of the participants were prominent figures in law firms, legal tech companies, and legal academia. They were asked to predict how legal AI might unfold in 2025, and to speculate about the “biggest surprise.” To identify common themes in their predictions, we used qualitative coding. Our frequency counts should be interpreted very cautiously given the open-ended question format and small, non-random sample. Moreover, most of our counts don’t lend to obvious interpretations. For example, on Question 1 (2025 predictions), some of the most common themes were: “innovation opportunity” (22), “regulatory challenges” (16), “access and equity” (11), “and ethical considerations” (10). We recommend reading the predictions yourself to get a better sense of these categories.
But there was one pattern we found especially meaningful: the sample was divided between 19 responses predicting a year of “gradual implementation” and nearly as many (16) emphasizing a year of “radical transformation.” These counts refer to Question 1, though it’s worth noting that Question 2 (the “biggest surprise”) tilted almost entirely toward radical transformation—with 16 responses coded as radical and only 1 as gradual. However, Question 2 may have primed participants to suggest something radical by asking for a superlative—the “biggest surprise.” So we think it’s more accurate to interpret the sample as roughly equally divided.
The gradualists expect “steady measured progress,” suggesting that “adaption to and adoption of this new technology will remain gradual” and “market penetration will remain modest.” They described continuing barriers to adoption, including the inherent conservatism of the legal profession around new technology. Others emphasized ethical concerns about generative AI, and suggested the profession will continue in their cautious “experimentation” phase.
Those in the “radical transformation” camp expect 2025 to bring legal AI’s first true “iPhone moment” or a “Netflix of legal,” where generative AI apps grow so popular they disrupt almost everything about how legal services are provided. This moment might be driven by overwhelming client demand, which some predict will explode over the coming months. It may also come from lawyers’ recognition of their own efficiency gains—the “rise of the x10 lawyer.”
Perhaps the most striking “radical” predictions related to agentic AI systems, which can make their own decisions and manage multi-step tasks (e.g., just three days ago OpenAI released its first agent, “Operator,” which can autonomously perform web-based tasks like booking travel and filling out forms). While every past human technology can be categorized as “tools,” an agentic AI may be more like a “decision-making partner” or an “AI co-worker,” as the survey participants noted. Some predicted 2025 will be the year lawyers begin “collaborating daily with AI colleagues” who have their own “distinct personalities, and specialized practice niches.” These agents could function as another lawyer in the office, attending meetings and offering suggestions at every step of legal workflow.
A related theme among those in the “radical” category was that law schools might “hit a tipping point” in 2025, as a few innovative leaders “revolutionize” curriculum, deeply integrating AI. As employers eagerly hire students with strong GenAI skills, a growing number of law schools might feel compelled to follow suit.
2025 could be a pivotal year for legal AI, though there is a notable divide about whether change will be revolutionary or evolutionary.
Empirical Studies of AI Lawyering (e-SAIL) 