Lawyers are training AI how to think like them
Summary
Legal professionals, including lawyers, judges, and paralegals, are increasingly taking on side jobs to train artificial intelligence models. Companies like Mercor and Micro1 hire these experts to perform tasks such as "red teaming"—creating complex legal scenarios to test model limits—and writing "golden responses," which serve as ideal examples for models to learn from.
This work addresses a significant bottleneck in legal AI development: unlike programming, legal reasoning is difficult to teach because much of the necessary knowledge is locked behind paywalls or confidential agreements. By providing professional judgment, these experts help ensure that chatbots used by millions of people provide accurate legal information.
For many practitioners, the motivation extends beyond the hourly pay, which typically ranges from $100 to $200. They view this work as a way to stay relevant in a transforming industry and to understand how AI will reshape the profession. While AI may automate routine tasks like contract drafting and research, experts believe it will struggle to replace high-level judgment, client reassurance, and courtroom advocacy.
(Source:Insider)