“#GeorgeSantos and these things we call #LargeLanguageModels actually have something in common, which is that they like to make stuff up.”

Nice one from @garymarcus on this week’s #LeShow with #HarryShearer.

(I can forgive the anthropomorphizing of ChatGPT by Prof Marcus, in this case. ChatGPT doesn’t have likes or dislikes. It does what its _designers_ like it to do: make shit up. We’re seeing _their_ preferences in this machine.)