OpenAI keeps moving the AGI finish line, and the market is starting to notice.
The Signal
The original OpenAI Charter defined AGI as "highly autonomous systems that outperform humans at most economically valuable work." Clean definition. Measurable. But watch what's happened since. As models got better at specific tasks, the company quietly shifted focus from capability benchmarks to something mushier: "systems that can reason and adapt like humans do." The goalpost didn't just move. It got wrapped in philosophical cotton.
This matters because the agent economy is being built on these definitions. Companies are raising billions, hiring thousands, and making decade-long bets based on when AGI arrives and what it can do. If the target keeps shifting, so does every business model downstream. We're already seeing it. Agents that can book your flights and summarize your emails are economically valuable. They're doing work humans used to do. But are they AGI? Depends who you ask and when you ask them.
The pattern is familiar. Web2 platforms redefined "privacy" until it meant nothing. Crypto projects redefined "decentralization" to fit whatever they were shipping. Now the biggest AI labs are redefining intelligence itself. The HN thread lit up because technical people smell the problem. If you can't pin down what you're building toward, you can't assess risk, you can't allocate capital intelligently, and you can't answer the question everyone is asking: how long until my job requires an agent license instead of a human?
The Implication
Stop waiting for AGI. Start tracking what agents can do quarter by quarter. Revenue per agent. Tasks automated. Jobs augmented versus jobs eliminated. The real timeline isn't when some lab declares victory. It's when your company runs leaner with agents than without them. That clock is already ticking.
Source: Hacker News Best