OpenAI’s president, Greg Brockman, recently suggested that its advanced model, known as GPT-6 Astra, could mark the arrival of something called artificial general intelligence, or AGI.
uncertain
model → suggest → something
If AI itself is often defined as the ability to simulate human intelligence, OpenAI defines AGI as an AI system that outperforms humans at most economically valuable work – to be better at the majority of digital and cognitive tasks a human can perform.
asserted
human → define → tasks
This includes tasks such as writing computer code, organisational activities or managing projects.
asserted
This → include → projects
For now, at least, it excludes work that involves physical labour, such as agriculture or construction.
asserted
that → exclude → agriculture
Though the integration of AI with robotics means such sectors of industry may not be the exclusive domain of humans forever.
uncertain
sectors → mean → humans
This definition by OpenAI focuses on breadth of competency by an AI system.
asserted
definition → focus → system
However, other experts maintain that AGI will be characterised by additional cognitive leaps beyond present-day AI models.
asserted
AGI → maintain → models
For example, Demis Hassabis, founder of Google DeepMind, has said that AGI should be able to do “pretty much any cognitive task that humans can do”.
asserted
humans → say → that
This definition makes AGI more about brilliance and a bit less about depth.
asserted
AGI → make → depth
Yann LeCun, former chief AI scientist at Meta, believes the term AGI itself is flawed and should be replaced by “human-level AI”.
asserted
AGI → believe → AI
These represent snapshots from a vigorous ongoing debate over when AGI will arrive and how to recognise it.
asserted
AGI → represent → it
Venturebeat magazine reported Brockman acknowledging this at a closed press briefing about GPT-6 Astra in September when he said: “When we started OpenAI, we kind of thought that there was going to be this well-defined moment that everyone would recognise … it hasn’t played out like that.”
asserted
it → report → that
But he added: “For me personally, I do think we’re there.”
asserted
we → add → me
Just a few days after Brockman made his comments, the chief executive of Anthropic, Dario Amodei, called in an essay titled: “We must pace the frontier” for the AI industry to slow down the pace of its development.
asserted
industry → make → development
Within hours of this essay being published, both Sam Altman and Elon Musk signalled their agreement.
asserted
Altman → publish → agreement
The evidence still points in more than one direction and is therefore inconclusive.
asserted
evidence → point → direction
The clearest shift with GPT-6 Astra is in how people use it.
asserted
people → use → it
Earlier models needed detailed instructions at each step.
asserted
models → need → step
Astra is built for longer jobs where the user sets a goal and some boundaries, and the model works out the sequence itself.
asserted
model → build → sequence
OpenAI’s demonstrations showed the frontier model handling several unrelated tasks at once.
asserted
model → show → tasks
The company presents this as moving past the pattern in which a human supplies the next instruction every time.
asserted
human → present → instruction
That is a real change, and it is the part most users will notice first.
asserted
users → notice → ?
One of the benchmarks show that Astra scored 99.9% on ARC-AGI-3, a test that places an AI agent in unfamiliar interactive games with no instructions and measures how efficiently it works out the rules.
asserted
it → show → rules
OpenAI estimates that the average human scored 48% when asked to work out the same unfamiliar games.
asserted
human → estimate → games
The detail matters here, however.
asserted
detail → matter → ?
ARC Prize, which runs the benchmark, reported that Astra’s best score through the shared “harness” it gives every model was 62.7%.
asserted
it → run → model
A harness is a kind of software operating system, or physical environment, within which the AI system works.
asserted
system → operate → which
ARC Prize provides Astra with this software harness.
asserted
Prize → provide → harness
The 99.9% came from a different harness, built by OpenAI for Astra, which preserves the AI model’s reasoning between requests.
asserted
which → come → requests
Essentially, Astra did better when it was able to reuse previous work and wasn’t “starting from scratch” between requests in terms of its reasoning.
asserted
it → do → reasoning
Evaluators disagree
asserted
Evaluators → disagree → ?
Independent evaluators disagree about how much of an advance GPT-6 Astra represents on its predecessor.
asserted
Astra → disagree → predecessor
The independent research institute Epoch AI places it right at the frontier, alongside other cutting-edge AI models such as Claude Fable 5.1, Kimi K3 and GPT-5.5 Pro.
asserted
institute → place → Fable
The independent analysis company Artificial Analysis did not rate it as being radically ahead of its predecessor – GPT-5.6 Sol.
asserted
company → rate → predecessor
Toby Walsh, an AI researcher at the University of New South Wales in Sydney, Australia, described the intelligence in frontier AI models such as GPT-6 Astra and others as “still today very jagged”, telling Al Jazeera that there are simple things that even the best AI models do poorly.
asserted
models → describe → that
On September 8 2026, OpenAI published a claimed solution to the Navier-Stokes problem, one of seven Millennium Prize problems in mathematics.
asserted
OpenAI → publish → mathematics
This has often been reported as Astra’s work.
asserted
This → report → work
OpenAI says the proof came from an unreleased internal model, significantly more capable than Astra, running roughly 10,000 AI agents (AI models that perform tasks in response to human instructions) in parallel for about 88 hours.
asserted
that → say → hours
Astra’s contribution was the formalisation and verification afterwards, which took a further 17 hours.
asserted
which → take → hours
The proof has not yet been checked by the mathematical community.
asserted
proof → check → community
…and 16 more, not listed.
The Aporia
