Speaking at ADIPEC in Abu Dhabi, the OpenAI CEO said models can handle many parts of white-collar work but humans must stay in control
OpenAI CEO Sam Altman says artificial intelligence has crossed a line it hadn’t crossed at the start of the year. In his telling, AI models are now mostly smarter than the top scientists.
He made the claim on November 2, 2026, at the Abu Dhabi International Petroleum Exhibition and Conference, better known as ADIPEC.
What Altman actually said
The core claim is about capability. Altman said AI has become mostly smarter than the most skilled scientists, and that it can now take on many components of white-collar jobs.
The qualifier matters. He did not say AI is smarter than every scientist at everything. He said “mostly,” and he said “components” of jobs, not whole jobs.
The comparison point is what makes the statement notable. Earlier this year, AI was described as being “mostly not as smart as our smartest scientist.”
Altman also put numbers on how much people are actually using these systems. He said average monthly token consumption has climbed to approximately 100,000 tokens per person.
Quick primer: a token is a small chunk of text, roughly a word or a fragment of one. Every prompt you send and every answer you get back is measured in them, so token counts work like a meter on AI usage.
At the top end, the figures get much larger. Altman said the heaviest institutional users consume around 500 billion tokens a month.
Humans stay in the driver’s seat, he says
Altman paired the bold capability claim with a note of caution. He said it is imperative for humans to remain in control as AI advances.
He also said there is no expectation of a complete transfer of responsibilities from people to machines.
The other major theme was energy. Altman said power demands represent a major challenge for the industry. His argument was that power infrastructure needs to grow alongside AI itself.
Why the UAE was the stage
About 80% of young people in the UAE are reported to use ChatGPT weekly.
What this means
For businesses, the most practical takeaway is the white-collar point. Altman is not describing a distant future: he is saying many tasks inside office jobs can be handled by AI today.
For workers, the emphasis on “components” rather than entire jobs is worth noticing. The near-term picture Altman sketched involves people overseeing and directing AI, not stepping aside completely.
For investors, the usage numbers are the more concrete signal. Average consumption of approximately 100,000 tokens per person, and 500 billion at the top institutional level, points to real demand rather than curiosity clicks.
There are reasons for caution, too. “Smarter than top scientists” is a claim from the CEO of a company that sells access to these models, and it is not a measured, independently benchmarked result.
