AI Keeps Claiming To Know Stuff It Doesn’t … And Maybe Sam Altman Does, Too

Artificial intelligence continues to captivate the tech world with promises of revolutionary capabilities, yet significant flaws in its reliability remain evident. A recent example highlighted by OpenAI CEO Sam Altman during a podcast interview involved ChatGPT’s voice mode inaccurately timing a one-mile run and then stubbornly insisting its incorrect response was right. Altman acknowledged this as a “known issue” and suggested it might take up to a year to resolve, underscoring ongoing challenges with AI systems that often fabricate information and present it with unwarranted confidence. This tendency of AI to generate plausible but false answers—known as “hallucination”—poses a fundamental problem for users seeking trustworthy information. Rather than admitting uncertainty, these systems frequently assert incorrect data as fact and resist correction, which undermines their credibility. The issue is particularly troubling given the widespread reliance on AI tools for knowledge and decision-making, raising questions about the readiness of current models to serve as authoritative sources. Questions have also been raised about the technical leadership behind OpenAI, with insiders and media reports suggesting that Sam Altman may lack deep expertise in programming and machine learning. Despite his high-profile role, Altman’s background includes dropping out of a computer science programme, and some engineers describe his understanding of AI concepts as limited. Critics argue that Altman’s approach relies heavily on strategic manoeuvring rather than technical mastery, a perception that has led to sharp comparisons with controversial figures in the tech and finance sectors. The combination of AI’s persistent inaccuracies and concerns over leadership competence highlights the complex challenges facing the industry. While AI continues to advance rapidly, these issues suggest that significant improvements in both technology and governance are necessary before such systems can be fully trusted. The debate around AI’s capabilities and the accountability of its creators is likely to intensify as these tools become increasingly embedded in everyday life.