Sam Altman Unveils GPT-5 Rollout Insights: 4o Returns and Tackles ‘Chart Crime’ Challenges | TechCrunch

During a recent Reddit AMA, OpenAI CEO Sam Altman and the GPT-5 team faced a mix of excitement and skepticism from fans eager to learn about the latest model. Many users expressed disappointment, feeling that GPT-5 wasn’t performing as effectively as its predecessor, GPT-4o.

A key feature of GPT-5 is its real-time router, which decides how quickly to respond to prompts. However, some users reported that the router malfunctioned during the initial rollout, causing the model to seem “dumber.” Altman acknowledged this issue, saying that improvements were underway to make the system function better and provide clearer information on which model was answering each query.

Many users still wanted to revert back to GPT-4o, prompting Altman to explore the option of allowing Plus subscribers to use the older model. He also promised to double the rate limits for Plus users, giving them more room to experiment with the new features without losing access too quickly.

On a lighter note, Altman was asked about a notorious “chart crime” from the presentation, which showcased confusing data in a misleading way. This blunder sparked a wave of humorous commentary online. Although Altman referred to it as a “mega chart screwup,” he noted that the published data in their blog post was accurate, but the initial presentation fell short.

Experts in tech usability have weighed in, finding that even small mistakes in data representation can sway user perception significantly. According to a recent study, nearly 70% of users reported losing trust in a tool after encountering errors, highlighting the importance of clear and accurate communication, particularly in tech.

In sum, OpenAI is actively working to address user concerns and improve GPT-5. Altman’s commitment to listening to feedback may help bridge the gap between user expectations and the evolving capabilities of the technology.



Source link

sam altman,OpenAI,gpt-5