ChatGPT told me about the time Don Rickles 'sent Lena Dunham a sext.' This makes me nervous about AI.
When ChatGPT is so constantly wrong about facts, what are we to believe? And how exactly are we supposed to "learn to use AI?"
TL;DR
- ChatGPT incorrectly identified Don Rickles as the male celebrity who allegedly sexted Lena Dunham in 2012.
- The author attempted to use ChatGPT as a super search engine to solve a blind item from Lena Dunham's memoir.
- AI can make factual errors, especially when information is scarce, by conflating different data points.
- The author questions the meaning and expected level of proficiency when told to "learn to use AI."
- The reliability of AI for factual information is a concern, particularly for mistakes that go unnoticed.