Sam Altman expects AGI system by end of 2026, prediction markets remain deeply skeptical

Sam Altman believes OpenAI will have a system it internally classifies as Artificial General Intelligence before 2026 ends. Prediction markets, which tend to be brutally honest aggregators of collective wisdom, are giving that outcome somewhere between a 9% and 15% chance of actually happening. What OpenAI is actually claiming In an interview with TIME magazine, Altman laid out OpenAI’s timeline with characteristic ambition. The company believes it can reach AGI, which it defines as systems capable of outperforming humans at economically significant tasks, by the close of this year. Altman was careful to note that OpenAI is “not quite yet” at AGI. OpenAI’s Chief Research Officer Mark Chen offered a more granular assessment, stating the organization is roughly “80 percent of the way” to achieving AGI. Co-founder Greg Brockman added his own framing, suggesting that the current moment might be retrospectively recognized as the period when AGI first emerged. The company has reportedly been developing its Astra model family, which has met internal benchmarks but hasn’t been publicly released. The definition problem Before anyone gets too excited or too dismissive, it’s worth unpacking w...