OpenAI Vice President of Product Peter Welinder assured that the company did not intentionally make the GPT-4 AI model any dumber than previous versions. It has been reported business insider.
ChatGPT Users of the GPT-4 model AI chatbot have been complaining for weeks that the new version performs worse, and some describe the changes as boring AI.
“No, we didn’t make GPT-4 stupid. On the contrary, we make each new version smarter than the previous one,” Welinder said.
According to him, such anger of users can be explained by the hypothesis that after a while they began to read the results of ChatGPT more carefully.
The top manager said, “When you use artificial intelligence more actively, you start to notice problems that you didn’t notice before.”
Several people responded to Welinder on Twitter and agreed that the new model has benefits. According to users, the chatbot began to perform tasks faster, but its results became noticeably worse.
On the OpenAI developer forums, many GPT-4 users expressed disappointment and posted examples of declining AI performance.
Previously on OpenAI in your name The timing of the release of artificial intelligence that poses a threat to humanity.