Maybe the ChatGPT implementation of GPT-4 is different than the one in Bing AI, but I tried to ask Bing AI to write a fairly simple Python-based ini-parser yesterday (and by that I really mean using the built-in configparser module), and while it got a good amount of the way there, but attempted to index a string with a string-key, which was weird. After multiple notices of this mistake, it produced something that _could_ work in some cases, but was definitely brittle.

> Maybe the ChatGPT implementation of GPT-4 is different than the one in Bing AI

Yeah I think it definitely is, but I don't know why. Bing is better at looking things up (perhaps unsurprisingly) but Chat4 is better at creating things.

I can confirm that GPT4 is much better than Bing on such tasks (have used both extensively for same prompts.)

Bing is backed up by GPT-4? No?

It is but they don't have to be exactly the same. Bing might be tuned for searching real time information and maybe cost less since at search engine scale is much higher (just a guess on my part).

My understanding is that they're similar but not the same. I think the rlhf process was different with GPT4 receiving much more human feedback.

I did an experiment of GPT-3.5 vs Bing vs GPT-4

Bing produced incomplete code.

Only GPT-4 was close enough to be correct.

Could be using different temperatures and prompts.