For all the intelligence that we like to ascribe to ChatGPT, the chatbot was essentially homeschooled. Its creator OpenAI trained it on the vast, imperfect glory of the public internet — one reason why ChatGPT makes so many embarrassing mistakes. A lawyer who recently used the chatbot to write his court brief realized he’d blundered when it cited six nonexistent cases.
How can ChatGPT get more accurate? Send it to college by training it on better-quality data.
That poses the tantalizing possibility of a new revenue stream for publishers and any other company that owns valuable, accurate text that could be used to train language models. ...
Learn more about Bloomberg Law or Log In to keep reading:
See Breaking News in Context
Bloomberg Law provides trusted coverage of current events enhanced with legal analysis.
Already a subscriber?
Log in to keep reading or access research tools and resources.