Same question. No precision. Are we sure AI is making us smarter?

CEO & co-founder @kramerapp.eth

FID 1,600,000,001

I asked “what is the exact number of published essays by Paul Graham on his website, paulgraham.com”. It ran some analysis and came up with 155.

I counted myself by copy pasting the list into a text editor and using the line numbers and got to 224.

When called out, ChatGPT tried again and returned 200+.

father, husband, css golfer, ux engineer, silently defining variables as /sdv sdv.wtf

I think for most queries, especially imprecise ones, it’s fine to lack precision.

Where it really shows generative AI’s weak points are questions like the infamous “how many r’s are in the word strawberry?” to which it almost never says three.

I think factual accuracy is more important to iron out than imprecision.