I normally never encounter any problems when asking it about errorcodes and showing my code for tips on what might be wrong. But this time was as bad as I have ever seen it. I’m using the GPT-4 model provided by the standard openai chat page. On two separate occasions, asking the same question about the same problem, I got answers that do not even come close to understanding the problem, much less solving it.
It didn’t even pick up the details from previous messages. It kept reiterating the same solution, even when making the changes and showing it the updated code, it completely ignored the updated code (which did not solve the issue) and kept stating the same problem, even though it wasnt an issue anymore. As I mentioned, this happened on two different occasions.
Mind you, the question was not about some complex coding problem, it was a simple compiling issue regarding package references in java.
I have been using GPT-3 since it came out, and I can vouch for it never being as bad as it is today.
Is there anyone else with a similar experience? I am aware of GPT drift, but this was a very drastic decline over a very short period of time. I also noticed a considerable lag in the responses, most probaly due to handling large workloads from other users.