Nvidia CEO Says AGI Has Arrived But the Concept Is Meaningless

Huang’s remarks came in response to questions about AGI, which he described as a milestone no longer useful for measuring progress. He said “for many tasks, we could say that we’ve already achieved AGI” and added that such milestones are “kind of senseless at this point.”

TechRadar’s article argues that AGI has become a malleable term that companies can redefine according to whatever their newest models happen to do well. It calls the concept a “promotion, not a scientific standard.”

The article dismisses apocalyptic scenarios of AI becoming self-aware and starting wars, but warns that AI still poses dangers through scams, voice and face cloning, phishing messages, and synthetic accounts on social media. These harms are “less theatrical” than nuclear war, but they are immediate and can trick people out of money or break into secure systems.

Huang suggested replacing AGI as a goal with questions about whether AI is doing “productive and useful work” and “generating profitable tokens.” TechRadar responds that profitability is not a safety test. It notes that scams are profitable, and that replacing human employees with an AI model that is 30 percent as good at 10 percent of the price is profitable in the short term — neither result is something to boast about.

TechRadar recommends shifting the industry’s focus from AGI to power and accountability: who uses AI, how it is used, and who is expected to clean up when it fails. Regulators should push AI companies to discuss liability, identity verification, ownership, and other legal details instead of chasing the perpetually delayed Skynet.