Among other “strange-sounding” proposals, OpenAI CEO Sam Altman stated in a recent piece on his personal blog that the business is open to a “compute budget” in order to “enable everyone on Earth to use a lot of AI” and guarantee that the advantages of the technology are shared widely.
The historical impact of technological advancement indicates that most of the metrics we care about (economic prosperity, health outcomes, etc.) improve over time and on average, but achieving greater equality does not appear to be determined by technology, and doing so may call for new ideas, Altman wrote. In particular, it appears that the capital-labor power balance could be quickly upset, necessitating early intervention.
Altman’s “compute budget” idea is one solution to this issue that might be simpler to suggest than to implement. The labor market is already being impacted by AI, leading to departmental downsizing and job losses. Experts have cautioned that, in the absence of appropriate government regulations and reskilling and upskilling initiatives, the emergence of AI technology could lead to widespread unemployment.
Altman has previously asserted that artificial general intelligence (AGI), which he describes as “[an AI] system that can tackle increasingly complex problems, at human level, in many fields,” is imminent. Altman cautions that regardless of its shape, an AGI won’t be flawless and could “require lots of human supervision and direction.”
Altman wrote, “[AGI systems] won’t have the biggest new ideas, and it will be great at great things but surprisingly bad at other things.”
However, Altman claimed that the true benefit of AGI would be realized when these systems are operated on a large scale. Like Dario Amodei, CEO of OpenAI rival Anthropic, Altman sees thousands or perhaps millions of extremely powerful AI systems performing tasks “in every field of knowledge work.”
It would seem that achieving such ambition would be costly. When it comes to AI performance, Altman did note that “you can spend arbitrary amounts of money and get continuous and predictable gains.” Perhaps for this reason, OpenAI has promised to invest up to $500 billion with partners on a massive data network and is apparently in negotiations to raise up to $40 billion in a funding round.
However, Altman also contends that the cost of using “a given level of AI” decreases by roughly ten times every year. In other words, while pushing the limits of AI technology won’t become less expensive, customers will have access to systems that develop more powerful over time.
DeepSeek, a Chinese AI startup, and others appear to support that idea with their capable, low-cost AI models. Although there is evidence that training and development expenses are also declining, Altman and Amodei have maintained that significant financial outlays will be necessary to reach artificial intelligence (AI) at the level of AGI and beyond.
If OpenAI can actually produce AGI-level systems, Altman stated that the business will probably make “some major decisions and limitations related to AGI safety that will be unpopular” in order to disclose these systems. In the past, OpenAI promised that, out of concern for safety, it would cease competing with and begin supporting any “value-aligned,” “safety-conscious” initiative that approaches developing AGI before time runs out.
That’s when OpenAI planned to stay a nonprofit, of course. The business is currently changing its corporate structure to a more conventional, profit-driven organization. By 2029, OpenAI apparently hopes to generate $100 billion in revenue, which is equivalent to the present yearly sales of Target and Nestle.
Given this, Altman went on to say that OpenAI’s objective as it develops increasingly potent AI will be to “trend more towards individual empowerment” while preventing “authoritarian governments from using AI to control their population through mass surveillance and loss of autonomy.” “I believe OpenAI has been on the wrong side of history when it comes to open-sourcing its technologies,” Altman recently stated. Despite its history of open-sourcing technology, OpenAI has typically preferred a proprietary, closed-source development methodology.
“We will expect everything to be smart; AI will permeate every aspect of the economy and society,” Altman stated. Many of us anticipate that we will need to grant individuals greater control over the technology than we have in the past, including greater open-sourcing, and acknowledge that there will need to be trade-offs made in order to strike a balance between safety and personal empowerment.
Ahead of this week’s AI Action Summit in Paris, Altman’s blog article has already encouraged several prominent figures in the tech industry to share their own ideas about the future of AI.
In a footnote, Altman stated that OpenAI does not intend to use the phrase artificial general intelligence (AGI) to terminate its partnership with Microsoft, a close partner and investor, anytime soon. According to reports, OpenAI and Microsoft established a contractual definition of artificial general intelligence (AGI) that, if fulfilled, would enable OpenAI to bargain for better investment terms: AI systems that can make $100 billion in profits. But OpenAI “fully expect[s] to be partnered with Microsoft for the long term,” Altman stated.
