OpenAI software is shit and they are hopeless at developing it

I've been using the GPT-4 case hearings module every day for the last 6 months. it is always getting worse and worse and worse. The "AI" side does not notice obvious flaws that I have explained to it multiple times. It's also completely incapable of taking into account any factors that are not written in the case summary. In other words, it will say that someone acted unreasonably because they [insert extremely ordinary thing] when the case summary does not say what their reason was for doing that thing. The AI will therefore say that they acted unreasonably without understanding that it doesn't know their reason.<br><br>It does not understand the questions anymore either. The questions are always in the same format, with the claimant in case A seeking judgment against the defendant in case B. This used to be "Claimant (Rswer: Defence) (Answer: Claim)". Now it is "Claimant (Rswer: Defence) (Answer: Defence)".<br><br>I'm not sure what the fuck is going on at OpenAI, but software development is clearly not their strong point.