For now, at least, it excludes work<\/a> that involves physical labour, such as agriculture or construction. Though the integration of AI with robotics means such sectors of industry may not be the exclusive domain of humans forever.<\/p>\n This definition by OpenAI focuses on breadth of competency by an AI system. However, other experts maintain that AGI will be characterised by additional cognitive leaps beyond present-day AI models. For example, Demis Hassabis, founder of Google DeepMind, has said<\/a> that AGI should be able to do \u201cpretty much any cognitive task that humans can do\u201d. <\/p>\n This definition makes AGI more about brilliance and a bit less about depth. Yann LeCun, former chief AI scientist at Meta, believes the term AGI itself is flawed<\/a> and should be replaced by \u201chuman-level AI\u201d.<\/p>\n These represent snapshots from a vigorous ongoing debate<\/a> over when AGI will arrive and how to recognise it. <\/p>\n Venturebeat magazine reported Brockman<\/a> acknowledging this at a closed press briefing about GPT-6 Astra in September when he said: \u201cWhen we started OpenAI, we kind of thought that there was going to be this well-defined moment that everyone would recognise \u2026 it hasn\u2019t played out like that.\u201d<\/p>\n But he added: \u201cFor me personally, I do think we\u2019re there.\u201d Just a few days after Brockman made his comments, the chief executive of Anthropic, Dario Amodei, called in an essay titled: \u201cWe must pace the frontier\u201d<\/a> for the AI industry to slow down the pace<\/a> of its development. Within hours of this essay being published, both Sam Altman and Elon Musk signalled their agreement.<\/p>\n So is Brockman right about AGI being here? The evidence still points in more than one direction and is therefore inconclusive.<\/p>\n The clearest shift with GPT-6 Astra<\/a> is in how people use it. Earlier models needed detailed instructions at each step. Astra is built for longer jobs where the user sets a goal and some boundaries, and the model works out the sequence itself. <\/p>\n OpenAI\u2019s demonstrations showed the frontier model handling several unrelated tasks at once. The company presents this as moving past the pattern in which a human supplies the next instruction every time.<\/p>\n That is a real change, and it is the part most users will notice first. <\/p>\n 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. <\/p>\n OpenAI estimates that the average human scored 48%<\/a> when asked to work out the same unfamiliar games. The detail matters here, however. ARC Prize, which runs the benchmark, reported that Astra\u2019s best score through the shared \u201charness\u201d<\/a> it gives every model was 62.7%. A harness is a kind of software operating system, or physical environment, within which the AI system works. ARC Prize provides Astra with this software harness.<\/p>\n The 99.9% came from a different harness, built by OpenAI for Astra, which preserves the AI model\u2019s reasoning between requests. Essentially, Astra did better when it was able to reuse previous work and wasn\u2019t \u201cstarting from scratch\u201d between requests in terms of its reasoning. <\/p>\n Independent evaluators disagree about how much<\/a> of an advance GPT-6 Astra represents on its predecessor. The independent research institute Epoch AI places it<\/a> right at the frontier, alongside other cutting-edge AI models such as Claude Fable 5.1, Kimi K3 and GPT-5.5 Pro. The independent analysis company Artificial Analysis did not rate it<\/a> as being radically ahead of its predecessor \u2013 GPT-5.6 Sol. <\/p>\n Toby Walsh, an AI researcher at the University of New South Wales in Sydney, Australia, described<\/a> the intelligence in frontier AI models such as GPT-6 Astra and others as \u201cstill today very jagged\u201d, telling Al Jazeera that there are simple things that even the best AI models do poorly.<\/p>\n On September 8 2026, OpenAI published a claimed solution to the Navier-Stokes problem<\/a>, one of seven Millennium Prize problems in mathematics. This has often been reported as Astra\u2019s work. It was not. 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. <\/p>\n Astra\u2019s contribution was the formalisation and verification afterwards, which took a further 17 hours. The proof has not yet been checked by the mathematical community. <\/p>\n Amodei\u2019s September essay<\/a> calling for an AI slowdown remains the strongest recent statement<\/a> about the risks of frontier models. He wrote that within six to 12 months<\/a>, a swarm of agents could be capable of taking over the internet with a persistent botnet. This could lead to the catastrophic failure of key systems and would cause damage possibly running into hundreds of billions of dollars. <\/p>\nEvaluators disagree<\/h2>\n
Claude misuse<\/h2>\n
![\"The](\"https:\/\/counter.theconversation.com\/content\/293600\/count.gif\")
<\/p>\n