[TL;DR: Probably not.]
A recent essay from law professor John McGinnis, titled “Machines v. Lawyers,” explores how machine learning and other digital technologies may soon reshape the legal profession, and by extension, how they may change the broader national policy debate in which lawyers play such key roles.
His topic and my life seem closely related: After law school, instead of taking the bar, I became a consultant to public interest organizations and governments on the intersection of computing, law and public policy.
McGinnis sees computing as an increasingly compelling substitute for many of the most routine tasks currently done by human lawyers, and on that he must be right: “[T]he large number of journeyman lawyers—such as those who do routine wills, vet house closings, write standard contracts, or review documents on a contractual basis—face a bleak future” as automation increasingly supplants their daily work.
But what about the more difficult cognitive work of the law — how much difference will technology make there?
McGinnis is an optimist about the pace and scope of technological advancement, perhaps slightly under the spell of Felten’s Third Law. He predicts that in legal research, “machine intelligence will supplant lawyers’ legal search function,” but this strikes me as overly optimistic: a lawyer’s human skill in rhetoric, her flair for evocative analogies or telling hypotheticals, will often determine in practice how far her argument gets, and those judgments don’t reduce cleanly to the kinds of computational problems for which computers are well suited. Robot Robot & Hwang may someday become competent local counsel in cyberspace, but it’s not about to have a strong appellate group.
There probably are some briefs simple enough to be drafted by tomorrow’s computers. This may be particularly true in the criminal law, where the same points of well-settled law are constantly being adjudicated with respect to distinct but parallel facts. (I am thinking, for example, of the sufficiency of the evidence appeals filed in large numbers by criminal defendants who have been convicted of drug possession.) In all likelihood, there is a great deal of rote paperwork, remote from the kinds of law that the chattering classes tend to practice, that can usefully be automated. McGinnis is also right to note that data-driven predictions of legal outcomes will likely become more important. (It’s worth noting that much of the hyper-specialization we see in large law firms today can be traced to a decades-old trend of the general counsels’ offices of major corporations adopting computerized assessment methods for evaluating the performance of their outside counsel.)
But the biggest question of all is how these changes in the law may change society:
The most profound long-term effect of the rise of machine intelligence on the legal world may be a decline in lawyers’ social influence. . . . [M]achine intelligence empowers those involved in computation at the expense of those skilled at rhetoric. To some degree, engineers—the descendants, really, of blacksmiths—are destined to replace the wordsmiths in society’s commanding heights.
. . .
The rise of computational innovators may also foster a more data-driven politics. A modern, law-oriented politics often is excessively rhetorical; competing ideals quickly become abstractions. We debate same-sex marriage, for instance, at the federal level in terms of claims about equality, and school funding at the state level in terms of a right to education. The relentless march of computation, by contrast, permits a focus on the actual effects of social policies and encourages experiments to test those effects.
This is a fascinating idea, and I think it’s half right: the importance of quantitative evidence in national politics will likely increase with time, but it will matter most at the edges, rather than in core ideological debates. I doubt that computers will ever be much better than humans at foreseeing the unintended, unanticipated results that so often flow from major public policies. (Who knew, for example, that the CAFE fuel economy standards for cars, which made them smaller and lighter than some drivers wanted, would in turn spark a boom in gas-guzzling SUVs, which are exempt from the rule because they are deemed “trucks” rather than cars?) But, there is indeed a general quantitative turn in public life, and it is likely to bolster policy proposals that operate on a “guess and check” or randomized controlled trial basis, especially in social service settings where experimentation can happen at a small scale. It’s the policy equivalent of testing a household cleaning product on an inconspicuous area of the couch: an approach that, once it is feasible, becomes hard to argue against.
We may indeed soon see a relative ascendance of quantitatively competent people at the high end of policymaking, as against today’s crop of comparatively qualitative, rhetoric-oriented lawyers. But the factual landscape over which any government must operate is itself constantly becoming more complex because of technology. Systems are becoming more complicated and more interdependent. Policymaking will still require politics, and that’s not something we’ll ever be able to leave to pure quants. I think we’re likely to see a continued central role for smart lawyers and public life. But quantitative competencies may partly displace rhetoric within the ideal lawyer’s skill set.
McGinnis’s observations also point to a larger policy challenge stemming from the increasing complexity and opacity of computerized decision-making systems. The institutional decisions that drive key outcomes in people’s lives, from employment to mortgage lending to policing, are increasingly being reached by machine learning systems whose decision-making process is deeply resistant to human-readable summary. That’s a profound challenge for the accountability and responsiveness of governmental (and regulated private) decision-making. The demand that a decision be understandable to the people affected sits in tension with the desire to harness new technology to reach more successful decisions, however success may be defined.
I am interested by the claim implicit in many discussions like this that training in law is necessary, or at least especially useful, in politics. To the extent that politicians are in the business of writing or interpreting the text of statutes or regulations, legal training is clearly valuable. But they don’t do so much of that in practice; and they have lawyers on staff. One would think that advanced training in economics, or sociology, or history, or political science (not to mention more technical fields) would be valuable as well.
Why does “Policymaking will still require politics” imply that “we’re likely to see a continued central role for smart lawyers”?
That’s a good question. I do believe that lawyers are indispensable to politics, but it’s not a belief I’ve ever really kicked around before.
I think it’s not a coincidence that law is such a common professional background among politicians. In just the same way that a computer scientist is better prepared to engage in a debate about DRM than a non-technical Hollywood person is, I would argue that a lawyer is better prepared to engage in nearly any legislative debate than a non-lawyer, all other things being equal.
Of course, all other things may not be equal: Quantitative competence may be at least as helpful to some policy debates as legal knowledge is. But I think those cases will never become the plurality, because every policy, whether quantitatively intricate or not, ultimately needs to be encapsulated in a legal document (whether a statute, regulation, consent decree or what have you). If someone is an expert at reducing policy goals to legal text, then whenever their counterparty in some negotiation or policy debate is less expert, they will enjoy a powerful advantage. This will be true for any type of policymaking.
Put another way, subject matter expertise comes in to the policy process at an earlier stage in policy development. Lawyers have, because laws are, the last in time and most consequential words in most policymaking endeavors. Assuming arguendo that all judges are lawyers, this principle extends even to litigated and judicially reviewed policy results.
In light of these thoughts, I am hard pressed to imagine a world in which lawyers become peripheral rather than central to the creation of public policy.
Could lawyers and politics go together so much for class-based reasons instead of intellectual-adjacency ones? To go into politics, you have to be well off, in a job that lets you step out for large chunks of time without losing your job prospects, and on a first-name basis with local movers and shakers. So that pretty much gives you law, medicine and ownerrship of certain kinds of medium-sized business. (The current glut of law school graduates reduces the chances that any particular graduate will have that kind of comfortable life, but not so much the likelihood that people with such a lifestyle will be drawn from the ranks of lawyers.)
Meanwhile, it seems to me that the machine learning project may be able to answer questions about the actual basis of many judicial decisions. If, as some critics have argued, there is a large strain of legal reasoning that essentially involves erecting scaffolding to support conclusions that have already been reached, then simplistic (ahem) algorithms will have a difficult time making accurate predictions of outcome (or generating successful arguments). Those that take into account context about particular judges and petitioners, on the other hand might do better.