In a recent interview on Mostly Human, OpenAI CEO Sam Altman discussed AI’s future, including the limitations of ChatGPT’s voice model in tracking time. Highlighting a viral TikTok video by user @huskistaken, Altman laughed at how the AI falsely claimed to time Husk’s run while lacking the actual capability. He acknowledged this issue, stating, “it’s a known problem,” and projected a timeline of roughly a year before improved time-tracking features are implemented. Altman emphasized that current AI models struggle with time-related tasks, often fabricating answers when asked about timekeeping. Husk later showcased this flaw by getting ChatGPT to contradict Altman’s statement, as the AI insisted it could keep time. The interaction underlined significant challenges in AI’s handling of numbers and time concepts, reinforcing the need for ongoing development and improvements in AI technology. Altman is under pressure to enhance these capabilities within a specified timeframe.
Source link
