Interesting comparison of ChatGPT and Claude, two generative “AI” assistants. Definite strengths and weaknesses to both; both appear to be fairly good at reasoning/logic, ChatGPT seems to be stronger at responding to mathematical and coding prompts, and Claude responds a bit more naturally to broader questions.

The bit where they were asked to summarize Lost made me laugh, too:

It appears that, like most human viewers of the show, both ChatGPT’s and Claude’s memory of Lost is hazy at best.

Last updated .