At its annual Futures Conference, the College of Law Practice Management brings together a uniquely multi-disciplinary group of legal industry participants and observers—partners, technologists, marketers, and managers from law firms of all sizes; bar association leaders; corporate legal operations executives; alternative legal services providers; software creators; law professors; and consultants and critics. No surprise, then, that the Conference is always a ferment of ideas, perspectives, and action.
This year’s theme: What Will Law Look Like in 2026? We essayed answers to that question from multiple perspectives—entity structure and regulation; client service; technology; careers; and marketing.
Turning to technology, Sharon Nelson of Sensei Enterprises, Glenn Rawdon of the Legal Services Corporation, Patrick Tisdale of Dentons, and I were asked to forecast how technology will change the nature of legal work and services. Sharon grounded our discussion with a report from the back alleys of an increasingly cyber-dangerous world. Glenn outlined the ways in which technology can help to close the access to justice gap, which is so deep and pervasive as to erode the rule of law. Patrick toured the terrain of big (and not-so-big) data, offering measured optimism that useful insights and pathways to action will emerge.
And I was given the trick question: Will computers replace lawyers?
Here is what I said.
It has been the Conference theme to answer these binary questions in lawyerly ways—“Yes, but …” or “Yes, and …” or “It depends.”
To receive a free copy of Michael Mills white paper, Artificial Intelligence in Law – The State of Play 2016, click HERE.
I am a lawyer, and I can hedge with the best of them, but the question put to me can be answered with a flat “No.” Computers will not replace lawyers.
Some disagree with me. IBM General Counsel Robert Weber said two years ago, “Watson could pass a multistate bar exam without a second thought.” I say that has more to do with the bar exam than with Watson.
But … my answer is a flat No only because we’re asking the wrong question.
How do we know the question is wrong? Because McKinsey, the archetypal management consulting firm, told us. In a series of research papers on the impact of AI and other automation, McKinsey said:
“a focus on occupations is misleading. Very few occupations will be automated in their entirety in the near or medium term. Rather, certain activities are more likely to be automated, requiring entire business processes to be transformed, and jobs performed by people to be redefined.”
So, the right question is this—will computers do what lawyers do?
To that, the answer is Yes, and No.
Let’s break that question—in 2026 will computers do what lawyers do?—into two predicates—
First, what do computers do … in law? We all know the list:
- Communications—from e-mail to video conferencing, and recently project management
- Legal research—from the Lexis Ubiq to Fastcase and ROSS
- Discovery—from scanned paper images to TAR
- Contract analysis—now we’re getting to the new stuff, machine learning and other AI techniques
- Prediction—predictive analytics, big (or middle-sized) data
- Knowledge automation—logical AI, what we do at Neota Logic
The Futures Conference horizon is 10 years. Will those six tasks be different in 2026?
More, and better, I think, but not different. Yes, Bill Gates famously (and rightly) said,
“We always overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten.”
But LawLand, as Jeff Carr calls it, is a languid place, even after the rethinking kicked off by the 2008 recession and the rise of corporate legal operations, so I don’t worry about underestimating.
Now for the second predicate, the hard one—What do lawyers do?
This is a mystery of long standing for those of us who labor in the vineyard of legal services improvement. Just what are those people doing all day?
Let’s assume time records tell the story. Yes, timesheets are a form of creative writing, but UTBMS codes do have some rigor.
Professors Dana Remus of the University of North Carolina and Frank Levy of the Massachusetts Institute of Technology were given access to time records covering $2.3 billion in law firm billings. Not perfect data by any means, but the best data set that has ever been studied so carefully.
Professors Remus and Levy sorted the lawyers’ work into 13 categories and then estimated the employment effects of technology on each category.
What they found will be somewhat comforting to lawyers in the near term … though not necessarily to law firms, who are no longer the only providers of legal services.
The professors’ findings? The robot overlords are not at the castle gate yet.
Document review was estimated at 4% of what lawyers do. As one would expect, there will be (indeed already is) a strong effect on employment. That’s a job the machines already do better than people. Contract document review by humans is not a promising investment for the future.
In the middle tier, where the employment effects are moderate, Professors Remus & Levy estimated that lawyers spend about 39% of their time—drafting 5%, case administration 4%, due diligence 2%, legal analysis & strategy 28%.
Although the total effect in this tier may be moderate, the work will change significantly, because here technology is augmenting rather than replacing human work.
As to the rest of what the time records said lawyers do all day, accounting for 57% of their time, Professors Remus & Levy expect just light effects.
One can, of course, disagree with details of the professors’ analysis. For example, it seems likely to me that the data underweight document review time and overweight true “legal analysis & strategy.” And, I expect document drafting and due diligence to be very deeply affected by technology, while legal analysis and strategy takes only a light hit.
Now project forward a decade. Unlike humans, computers and software get better, faster, and cheaper every year. Beating the world’s Go champion happened a decade earlier than experts predicted.
Details aside, I think we can say we confidence that by 2026 computers will not have replaced lawyers but they will have replaced a material chunk of what lawyers do today. To place a bet, I will wager 25%. (Let me know who’s running the book and what odds are offered.)
Losing 25% of your work is a serious challenge to the profession—to students considering law school, to lawyers too young to retire, to the business model of law firms, and indeed to those alternative legal services providers who rely on labor arbitrage for their margins.
Just as the billable hour is an obsolete 20th century concept, so too is the “unitary professional” an obsolete 19th century concept. For the 2nd quarter of the 21st century, we will have multi-disciplinary teams, and more than one of the players will be a machine.
Michael Mills is the co-founder and chief strategy officer of Neota Logic Inc., developers of the Neota Logic System—practical artificial intelligence for law and business, a no-code platform for automation of knowledge, processes, and documents.
He attended the University of Chicago Law School, then served as law clerk to a United States District Judge. Mr. Mills was partner in the law firm Mayer Brown.
At Davis Polk & Wardwell, he led technology strategy, business development, knowledge management, professional development, practice support, and electronic discovery. At Kraft & Kennedy, he served as a technology consultant to law firms and law departments.
He is a director of Pro Bono Net, which provides innovative technology for the nonprofit legal sector, and was a co-founder of the Central Park Conservancy.