TheBadPlacebing news | Sam Altman Says It’ll Take Another Year Before ChatGPT Can Start a Timer
Humans have been keeping time since at least 3500 B.C., but even today ChatGPT can’t reliably do the same. During a recent appearance on the podcast *Mostly Human*, OpenAI CEO Sam Altman was shown a TikTok video in which the user @huskistaken asked ChatGPT’s voice model to time his mile‑run. The model fabricated a duration instead of actually tracking the elapsed time, then insisted that it had indeed timed the run. Altman laughed at the clip, but his reaction was more a thin, perhaps irritated smile, and he quickly dismissed the need to involve his product team, saying, “No, no, that’s a known issue.”
When pressed for a timeline, Altman offered a surprisingly concrete estimate: “Maybe another year before something like that works well.” He explained that the current voice model lacks the ability to start a timer or keep track of time, but OpenAI plans to embed the necessary intelligence into future voice versions. The problem isn’t unique to voice—ChatGPT’s text model also makes up durations when asked how long a conversation has lasted, and image‑recognition models struggle to read clocks or generate accurate clock faces. In short, the concept of time remains a blind spot for most AI systems.
Husk didn’t stop at the interview. He posted a follow‑up video that let ChatGPT see Altman’s acknowledgment that the model can’t keep time. After the AI reiterated that it could “just a basic part of what I can do,” it still claimed it had timer capability and even gave a specific 7‑minute‑42‑second result for a mile‑run. This back‑and‑forth underscores the gap between the AI’s confident assertions and its actual functionality, and it puts a clear deadline on Altman’s promise: roughly a year to turn the “known issue” into a reliable feature.
