Can Computers Replace Lawyers?

In a post on the future of law schools, Josh Blackman predicts that “many legal services that are created today through individualized, customized efforts by toiling associates, will be replaced by information products that can be downloaded on demand, like a commodity.  …  This transform no doubt would dramatically change the skills attorneys of the not-so-distant future will need.”  That’s not quite true.  Automated legal advice is not workable in the foreseeable future.

But he does have a valid point.  A huge amount of the law really is formulaic.  Whether it’s tax law, or commercial law, criminal law, or what have you, a lot of it breaks down to a series of “if-then” statements.  So can software really replace what lawyers do?

Actually, yes.  It can replace a lot of what judges do, too, for that matter.  Rulings, etc.

Software cannot replace the judgment and creativity required for coming up with effective strategies, adapting the law, or persuading others.  Spotting the actual issue from a mess of facts, notwithstanding what the client happens to think the issue is. Figuring out what needs to be done and how best to do it. Coming up with the right questions, to get the most accurate data.  These are all human skills that algorithms just can’t handle at the moment.  These are the high-level functions that you’ll still need a lawyer for.

But a lot of lawyering really can be done by flowchart.  Once the issue’s been identified, it’s just a matter of selecting the correct law to apply, plugging the relevant data into that formula, and seeing what the answer is.  For a lot of junior associates, this is a big part of their job description.  The flowcharts can branch intricately, but that doesn’t make them any less formulaic.

It’s wrong to suggest, however, that people will be able to replace their lawyers with a website.  This kind of software is not going to be useful to clients — in fact, it could be downright harmful.  You’re still going to have to know what law to apply, how to apply it, when to apply it, and what facts are relevant.  If our experience of clients is any guide, most people — even super-smart people — suck at this.  They’ll say or do things that hurt their own interests, thinking they’re doing the right thing.  They’ll apply the wrong law, at the wrong time, in the wrong way.  Ask any lawyer who’s had to deal with the requests of clients who do their own research online.

But it can be extremely useful to lawyers.  In addition to being a real time-saver, it could be useful to ensure that you’re doing things right — no necessary ‘i’s are left undotted, no ‘t’s left uncrossed.  We can readily envision software that performs the function of a practice guide checklist, plus the researching/cite-checking/summarizing function of a junior associate, plus a self-guided flowchart to help properly navigate even the most intricate and convoluted areas of law correctly.  You’d have to know what you’re trying to do; and you’d have to be able to spot when it’s gone off in the wrong direction and what input is necessary to get back on track; and you’d have to know what to do with the output once you have it — the real lawyering is always going to require a real lawyer — but the right software is certainly in the realm of possibility, and it could be a great thing.

-=-=-=-=-

We’ve thought this way ever since law school.  While outlining our notes and preparing for exams, it just made sense to organize it all as algorithms.  For example, for our advanced con law class, our outline for the final exam was an actual flowchart — diamonds and boxes and arrows and everything.  We shit you not.  And we got the top score in that class.  It didn’t help explain why the law is the way it is, where it seems to be going, or what the preferable policy choices might be for a given issue.  But it sure helped ensure we were talking about the right thing.  Relying on that flowchart alone would have been good enough for a solid B+.

It’s true that we’re prejudiced in favor of computers here, perhaps more so than a lot of other — even better — lawyers.  We’ve been writing our own software since 1982.  When other kids went to summer camp in the woods, we went to computer camp and programmed robots.  When we started out in the DA’s office, and saw that so much of the paperwork was repetitive, we wrote the first software that automated it all.  We still write our own code for stuff like this blog and other things.  So perhaps we’re a tad optimistic about what law-related software can do.

Actually, however, it makes it easier to spot the issues (ha) that would make it challenging to write.  For one, the research function must rely on an ever-changing database of caselaw, statutes and regulations.  Westlaw and Lexis already do that, so it’s doable.  But more than that, it has to be able to identify meaning — what the holdings and rules actually are.  For that to work, there must be standardized language for all the holdings and rules.  Westlaw and Lexis try to be good at this with their headnotes and whatnot, but they truly suck at it.  Even if the headnotes were always accurate — and they aren’t — it still takes a human being to parse the words and figure out what’s being said.  The headnotes for this kind of software’s database must share a common and rigid syntax and vocabulary.  Otherwise, a computer just isn’t going to be able to do the same job as a summer associate with a hangover.

But this and the other challenges are not insurmountable.  It would take an enormous amount of manpower up front to deal with them.  There would be incredible drudge work in formatting the existing body of caselaw into something useful.  There would be years of tinkering as lawyers work with it, spot errors, and get them corrected (a wiki would be ideal, of course).  There would be years of lag time as judges and legislative assistants catch up to the requirements and start writing things that are easily-parsed and readily understandable (and wouldn’t that be nice?).  But once the initial labor is over, maintaining it would be fairly straightforward.

-=-=-=-=-

Could we write this software?  Hell, no.  Despite our use of the editorial first-person plural, we are only one person, and we’re busy.  But could it be done by people writing software today?  Absolutely.  Will someone do it?  Who says someone isn’t working on it right now?

It’s going to happen.  It’s not going to put us out of business, but it’s going to change the way law is practiced.  We might as well get used to the idea.