I had been writing about ongoing issues with chatGPT. Apparently I can’t post more than 3 times because I am new - but it’s been weeks since. So, here I am again actually trying to help improve things by drawing attention to glaring issues.
GPT4 has lost what was once a not bad reasoning capability. I will leave this here. This was a test following something it really messed up on involving chemistry. Something that a naive user might have followed and gotten into serious trouble as a result. This is about where 3.5 was back in April.
It’s useless in this state. It cannot be trusted. For god’s sake will someone in the company fix these issues or acknowledge they exist. This is turning into a very expensive waste of time lab experiment.