During a recent Reddit AMA, OpenAI CEO Sam Altman acknowledged that the rollout of GPT-5 had been rocky due to troublesome technical issues and user concerns. He characterized one of the most publicized misunderstandings of the rollout as the infamous "chart crime," with the "chart crime" beginning with a troublesome bar graph from the GPT-5 demo that incorrectly interpreted their performance data. Altman referred to it as a "mega chart screwup," and he apologized for the problem but noted that the charts have been updated; current charts are included on the OpenAI blog.
User feedback quickly confirmed that GPT-5 had improved reasoning and capability, but it did not have the familiarity and conversational tone that consumers came to appreciate when using GPT-4o. In fact, users reported diminished performance and emotional appeal to GPT-5, and some even canceled subscriptions. Therefore, OpenAI decided to offer Plus users access to GPT-4o again, as familiarity and personality are important to the user.
These events should remind us that technical proficiency, is important, but AI must also feel that it is genuinely helpful and emotionally appealing to the user. Thus, the combined feedback - correcting visual representations and giving back the user's AI relative personality - suggests that OpenAI was trying to balance new technology projects with user experience and trust.
