Is ChatGPT AGI?
Why ChatGPT (and its peers) aren't AGI yet
AGI, in the serious sense, is AI that can do essentially any cognitive task a skilled human can — including working autonomously and reliably end-to-end, the "drop-in remote worker." Frontier chatbots clear a lot of the capability bar but miss the autonomy one: they assist brilliantly within a task, but don't own an entire job unsupervised without human cleanup. That gap is the whole ballgame.
| AGI test | Frontier models, mid-2026 |
|---|---|
| Match skilled humans on scoped tasks | Largely yes (~83% GDPval) |
| Strong agentic / coding work | Yes (~80% SWE-Bench Pro) |
| Reliable unsupervised, end-to-end ownership | No |
| Autonomously conduct AI research | Undemonstrated |
Is it superintelligence? Not remotely
Superintelligence (ASI) means far exceeding the best humans at essentially everything — a bar today's models are nowhere near. If anything, the honest framing is that current systems sit in a powerful "very capable assistant" band: past a strong graduate on many scoped tasks, short of a reliable autonomous worker, and light-years from ASI. See AGI vs superintelligence for where each line sits.
So how close is it?
Closer than skeptics say, further than the "AGI is here" headlines claim. The decisive milestone — autonomous AI research — hasn't been demonstrated, which is why the headline AGI-by-2027 prediction resolves by January 2028 rather than being settled now. Watch the single number that tracks it move in real time on the Thesis Tracker.
Frequently asked questions
No. As of mid-2026, ChatGPT and other frontier models reach near skilled-human performance on scoped tasks (~83% GDPval, ~80% SWE-Bench Pro) but lack reliable, unsupervised, end-to-end autonomy — the bar serious AGI definitions require. They are also nowhere near superintelligence.
No frontier model in 2026 meets the serious AGI bar. They are extremely capable assistants that clear much of the capability threshold but miss the autonomy one — no system reliably owns an entire job, or conducts AI research, without human supervision.
Reliable, accountable, long-horizon autonomy: doing an entire job end-to-end without human cleanup, and ultimately conducting AI research itself. That is the milestone behind the AGI-by-2027 verdict, currently graded Open and resolving by January 2028.
Not remotely. Superintelligence means far exceeding the best humans at essentially everything. Current models sit in a 'very capable assistant' band — strong on scoped tasks, short of autonomous work, and far from superintelligence.
The live scorecard updates as models ship and verdicts change.
View the live scorecard →Get the weekly AGI progress briefing
We track the exact line between "very capable assistant" and real AGI. Subscribe to hear when it moves. Free, no hype.
Subscribe free →