[Ars Technica] - Study claims ChatGPT is losing capability, but some experts aren’t convinced
[Ars Technica] - Study claims ChatGPT is losing capability, but some experts aren’t convinced
Either way, experts think OpenAI should be less opaque about its AI model architecture.
From my own experience chat gpt has regressed in the answers I get. Bard has actually started to give me answers that are more usable.
I have yet to try bard but yes. ChatGPT has declined significantly in useful answers since the launch of 4. It makes me question the value of their model at this point. I’m personally excited to deploy and use these open source ones such as falcon.
Llama2 just came out and it's very good.
Some experts are morons
Current hypothesis: When you use it more heavily, you start noticing issues you didn't see before."
Seems pretty sus to me. I use it pretty rarely and yet the decline in performance has been clear as light to me. Before: I ask it about a common library, it answers perfectly. Now: I ask it about an even more common library, it hallucinates and it turns out SO is more helpful than GPT which is actually quite surprising. In January I was able to teach myself typescript with GPT as my sole teacher. Now I can't get a basic answer about the datetime library. lol.