Early last year I put ChatGPT to a test to determine the artificial intelligence gizmo’s acumen for insurance coverage. It scored an “F” – getting 4 out of 5 questions dead wrong.
I decided to give it a re-test, including signing up (and paying) for the most up-to-date version. The AI technology was much improved. I gave it a “C.” While the advancement was dramatic, it still got some things very wrong. In addition, there were instances where it had a correct answer and should have quit while it was ahead. Instead, it went on to add some unnecessary commentary, which introduced vagueness and uncertainty to what could have been a much more accurate answer.
Here is my lengthy and detailed report card on ChatGPT’s insurance coverage re-test:
https://www.coverageopinions.info/Vol13Issue1/ChatGPT.html
|