By 2017, if Moore’s Law continues to hold, computers will have the processing power of a human mind. This is not science fiction. It will happen, and it means that sometime during this century, computers will be able to at least simulate human thought, if not think independently. Will that mean computers will be able to replace lawyers?
Let’s back up.
Full disclosure: Clio was so eager to have me at its conference that it paid for my flight and hotel, then plied me with fancy cocktails at Zed451, a trendy Chicago bar with a Canadian name (although everyone from so-charmingly-Canadian-it-hurts-eh Clio acted really surprised when I brought this up).
- 1989 — Garry Kasparov easily defeats “Deep Thought,” designed by computer science student, Feng-hsiung Hsu, in a two-game chess match.
- 1996 — Eight years later, Garry Kasparov loses once to Feng-hsiung Hsu’s next chess-playing computer, “Deep Blue,” developed at IBM. But Kasparov goes on to win the match.
- 1997 — Only a year later, Kasparov beats Deep Blue in the first of six games. This is the last time a human beats a supercomputer at chess. Deep Blue goes on to win the match.
- 2011 — “Watson,” also developed by IBM, wipes the floor with Jeopardy champions Ken Jennings and Brad Rutter.
- 2013 — The Urban Dictionary has to be removed from Watson’s memory after it starts swearing at everyone. (Okay, Ed didn’t say anything about this. It’s just awesome.)
So what are the computers that are supposed to eventually replace lawyers’ jobs? Are they more like Deep Thought or Watson? Ed suggested they are on the Deep Thought end of the spectrum. I think they more like this:
But eventually — this seems inevitable — legal computers will reach something like the level of Deep Blue relative to Kasparov, or Watson relative to Ken Jennings. It may happen in ten years or thirty, but it will happen, and it will probably happen in this century.
Don’t worry too much about your job, though. Computers smarter than humans will quickly produce computers smarter than they are, and so on, probably faster than we can predict. When this happens, the technological development curve goes close to vertical, and we can no more predict what happens on the other side of that vertical line than ants could have predicted the Industrial Revolution.
It’s quite possible that the Singularity — that vertical line — will signify the end of lawyers. It’s just as likely that we will all turn into human batteries or that computers will become our benevolent caretakers and lawyers won’t be necessary in our post-scarcity society.
The best news may be that the Singularity marks the moment legal futurists are officially — as Watson might have put it until he was censored — full of shit. All bets are off.
Ed Walters, to bring this around to the point, is not a legal futurist. He runs Fastcase, and his job is to think his way around the manpower Wexis can bring to bear on legal research. Ed is sitting on a pile of “big data” — a pretty-comprehensive body of cases, statutes, and other authorities — and imagining awesome things to do with it.
For example, the Bad Law Bot flags cases with negative citation history, generally as good or better than human editors. Talking with Ed makes my head spin with all the amazing possibilities for that data and Fastcase’s computing power. Wouldn’t it be awesome to just upload your opponent’s brief and get an annotated research profile, including the key cases and the most-relevant cases he didn’t cite? Or, like what Lex Machina is actually already doing for IP cases, build a profile of how a judge is likely to decide an issue based on their previous decisions:
As cool as these things are, they are Deep Thought–level problems. Watson-level problems, like arguing a motion or making an opening statement, are still a ways off. But maybe not that far off. There were eight years between Kasparov’s rout of Deep Thought and his victory over Deep Blue. But a year later, Deep Blue took the lead and never gave it back.
Two years before a computer becomes capable of defeating a lawyer in court, or of conducting a trial as well as a judge, it may look as inept as Deep Blue at its first match against Kasparov. Two years later, it will do your job better than you.
It looks like you’ve got at least 4 or 5 years to find another job, but probably not much more than 15 or 20.