Just got access to GPT4 via ChatGPT and it's doing shockingly (and creepily) well on my test of "Who is X?" - here's its answers for Simon Willison and for Andy Baio
Both of them appear to be entirely accurate - in the past I've seen all sorts of wild hallucinations from this kind of prompt
@simon How well can it code compared to 3?
@matrix Trying that now and the results are pretty extraordinary, it seems to be able to write Datasette plugins from scratch
@matrix I think both!