OpenAI, the leading artificial intelligence (AI) company, has been facing complaints from users about the performance of their latest AI model, GPT-4. However, OpenAI’s executive, Peter Welinder, has responded to these claims, stating that they have not made GPT-4 dumber. In fact, he mentioned that each new version of the AI model is designed to be smarter than the previous one.
The concerns raised by users were highlighted in a recent report by Insider, with some users describing GPT-4 as lazier and dumber. Welinder took to Twitter to address these complaints, suggesting that the observed issues might be more noticeable because the system is being used more heavily.
Welinder encouraged users to share specific examples where they believe GPT-4 has regressed in performance. Twitter users responded by sharing instances and examples directly with him.
One user, Ethan Harris, a retired Army medic, replied to Welinder’s tweet with a specific example. Additionally, another user pointed to a blog post by Nihit Desai, a startup co-founder, who reported degraded performance for data labeling tasks. OpenAI’s own developer forums were also highlighted, where numerous users have expressed frustration and posted examples of GPT-4’s inferior performance.
Insider had contacted OpenAI for comment prior to publishing their story, but the company did not provide a response.
OpenAI’s claims that GPT-4 has not become dumber appear to contradict the concerns raised by users. The reported performance degradation has led to disappointment and frustration among some users who rely on the AI model.
Despite OpenAI’s statement, the evidence provided by users through specific examples suggests a decline in GPT-4’s performance. It remains to be seen how OpenAI will address these concerns and whether they will take steps to improve the AI model based on the user feedback received.
As of now, the issues surrounding GPT-4’s performance have not been fully resolved, and users are eagerly awaiting OpenAI’s response to the complaints. Only time will tell if OpenAI can rectify the situation and meet the expectations of their users with future AI models.
