Are You Ready for the AI Apocalypse?

Just today, two federal judges--a Reagan appointee and a Biden appointee--were caught pretty obviously using AI to write opinions. They were caught because they referenced nonexistent parties to the suits, nonexistent opinions, completely misrepresented existing opinions, and the like.

Part of the issue, I'm persuaded, is that at its root, AI works on the same kind of regressions that quality engineers and such have been using ever since Deming and Shewhart got started, and with the typical p values used, you come to the wrong answer some of the time. Now multiply that by the dozens/hundreds of citations and conclusions in a typical legal decision, and you're pretty much guaranteed to make some hilarious mistakes.

As if the judge was on LSD, which is why we call this effect a "hallucination". Now more than ever, the ability to think for yourself and discern good evidence from garbage is huge.