Rob reported 4 different legal cases where lawyers quoted fake cases or other hallucinations that didn’t pass any sanity check.
馃幆 If you plan to use LLMs for your defense, think twice.
馃幆 If you seek lawyers based on a ChatGPT search, you may end up with one of Freund’s cases.
We know firsthand how hard some teams work – Angel Reyes & Associates being one notable example – but each market is crowded with “experts”, whereas some are dramatically better and others may be outright dangerous (or put you in jail).
Harvey AI is the most notable startup in this space, raising another $150M this week. And unlike most startups in the space targeting consumers directly, Winston Weinberg, Harvey’s CEO, clearly looks at law firms as clients.
Professional solutions in the space empower experts instead of selling snake oil promising miracles.
馃帗 After spending many thousands of hours in deep integrations with both LLMs and ML models here, it’s clear that certain cases work “well enough”, but many aren’t production-ready, need red teams to perform extra sanity checks, and still require final human oversight on top. Most of our solutions are deterministic with creative enrichment when it makes sense.
As for lawyers abusing ChatGPT – we may hear about disbarments in the foreseeable future.
