Minimal News_

With the release of OpenAI’s latest model o3, there is renewed debate about whether Artificial General Intelligence has already been achieved. The standard skeptic’s response to this is that there is no consensus on the definition of AGI. That is true, but misses the point — if AGI is such a momentous milestone, shouldn’t it be obvious when it has been built?

In this essay, we argue that AGI is not a milestone. It does not represent a discontinuity in the properties or impacts of AI systems. If a company declares that it has built AGI, based on whatever definition, it is not an actionable event. It will have no implications for businesses, developers, policymakers, or safety.