Today’s chess-playing computers can crush the best human players without breaking a sweat. This wasn’t always true. A couple of decades ago teams of the strongest humans and the most powerful computers were stronger than either humans or computers alone. These teams were sometimes called “centaurs.” They combined the strength of a mighty beast with human judgment.
For at least the next few years, legal centaur teams combining the experience of the best lawyers and top AI apps will always win over either human lawyers or AI apps working alone.
Today’s best legal AI experts (including Richard Susskind) believe that this may not always be true. They speculate that eventually computers will reach a stage of “hyperintelligence” in which AI systems become unfathomably more capable than humans. We are not there yet, and we may never get there. For the foreseeable future, experienced lawyers who know how to use AI will dominate.
Today, I have no problem asking an AI app a simple question about state licensing of music therapists. I would verify its analysis before relying on it for anything important, but AI is now my first choice for a relatively simple question where the stakes are low.
I would never dream of relying on an unassisted, unsupervised AI app for an important issue in $40 million litigation.
At the same time, today I no longer rely solely on my unassisted human judgment on a high-stakes matter.
The rules of thumb I explained in a January LLRX.com article describe the best approach:
- Never rely on anything AI tells you about crucial issues.
- Always ask AI for advice on crucial issues.