Sam Altman has given his most audacious forecast yet regarding when AI will outperform human intelligence.
Altman, the AI optimist, has previously touted AI’s ability to outthink humans, but the technology’s advancement during the last three years since OpenAI launched the AI arms race with the publication of ChatGPT has showed him that this moment may arrive sooner rather than later.
In a German media interview, he stated, “I would be very surprised if we don’t have extraordinarily capable models that do things that we ourselves cannot do by the end of this decade, by 2030.”
According to Altman, the company’s most recent AI model is already smarter than he is, and things appear to be getting better in the future.
Even while AI would still struggle with certain tasks that humans can complete with ease, Altman predicted that models created as early as 2026 might be “quite surprising” and advance quickly.
In the not-too-distant future, he added, he can readily envision a world in which artificial intelligence (AI) does between 30 and 40 percent of the jobs carried out in the economy today.
Altman’s forecast is a little less hopeful than that of other CEOs and AI experts who have predicted that the technology would advance significantly in the coming years. Dario Amodei, the CEO of Anthropic and a former employee of OpenAI, is notable for his prediction that by 2027, AI would surpass humans “in almost everything.” Elon Musk, the founder of xAI and a man known for his optimistic forecasts, thinks AI will outsmart the most intelligent person by next year.
Even with the capabilities of advanced technology, Altman pointed out that people would always be concerned about what other people are doing.
These attributes, in his opinion, will become more and more crucial in the field of artificial intelligence. “Even though we will have a fantastic tool at our disposal, we still need to determine what we should do, what other people want, and what other people will find useful.”
Exceeding the boundaries
As the fifth most popular website globally, according to Altman, ChatGPT is undoubtedly constrained by the significant energy consumption and enormous data centers required for its operation. Altman previously stated that although the business has developed models that are more sophisticated than GPT-5, it is unable to make them available to the public due to a lack of the necessary infrastructure.
Altman remains hopeful. OpenAI, Oracle, and SoftBank are quickly expanding an 800-acre data center facility in Abilene, Texas, which will be the centerpiece of the Stargate infrastructure project, which Trump has authorized. Over the next years, OpenAI and Oracle plan to construct five additional data center complexes in the United States, according to Fortune’s Sharon Goldman.
This site represents a mere portion of our ongoing construction. At a gathering at the Texas location this week, he stated that all of that would still not be sufficient to meet ChatGPT’s demand.
All of this infrastructure will help AI achieve its ambitious goal of changing the world, but Altman believes the technology is well on its way.
He told Die Welt that it will be extremely likely in a few more years for AI to produce scientific breakthroughs that humans are unable to accomplish on their own. In his opinion, that will begin to resemble what we may appropriately refer to as superintelligence.
