February 10, in a new post on the personal blog ofOpenAI CEO Sam.AltmanA number of "strange-sounding" ideas have been proposed, including a "computational budget" to "make AI widely available to everyone on the planet" and to ensure that the benefits of the technology are widely distributed. and ensure that the benefits of the technology are widely distributed.
1AINote that Altman points out that the historical impact of technological progress suggests that most of the indicators we care about (e.g., health outcomes, economic prosperity, etc.) generally get better in the long run and in an average sense, but that increased equality does not seem to be an inevitable consequence of technological progress, and that new ways of thinking may be needed to achieve this. In particular, he mentioned that the balance of power between capital and labor could be easily upset, which might require early intervention.
However, solutions to this problem, such as Altmann's concept of "computational budgeting", may be easier to propose than to implement. Already, artificial intelligence is affecting the labor market, leading to layoffs and departmental downsizing. Experts warn thatWithout proper government policies and retraining and upskilling programs, the rise of AI technology could lead to mass unemployment.
This is not the first time Altman has claimed that artificial general intelligence (AGI) is coming. He defines AGI as "artificial intelligence systems capable of solving increasingly complex problems on a human level in a wide range of domains". Altman warns, however, thatRegardless of the form of AGI, it will not be perfect and may "require significant human oversight and guidance."
"AGI systems won't generate the biggest new ideas," Altman writes, "they'll excel in some ways, but may be surprisingly bad in others." But the real value of AGI, he argues, will come from running these systems at scale. Like OpenAI competitor Anthropic CEO Dario Amodei.Altman imagines thousands or even millions of super-powerful AI systems handling tasks in "every area of knowledge work."
One might think that realizing this vision would be costly. In fact, Altman observes, "you can invest as much money as you want to get a sustained and predictable increase in AI performance." That may be one of the reasons OpenAI is rumored to be in talks for a round of funding of up to $40 billion and promises to spend up to $500 billion with partners on a vast network of data. HoweverAltman also noted that the cost of using "a certain level of AI" decreases to 1/10th of a percent approximately every 12 months. . In other words, while the boundaries that drive AI technology won't get any cheaper, users will have access to increasingly powerful systems.
Powerful and inexpensive AI models from companies like Chinese AI startup DeepSeek seem to support this view. There is evidence that the cost of training and developing AI is also falling, but both Altman and Amodei argue that huge investments will be needed to achieve AGI-level AI and beyond.
As for how OpenAI plans to release an AGI-level system (assuming the company can actually create such a system).Altman said the company may make some "significant decisions and restrictions related to AGI security that may be unpopular"OpenAI has promised to stop competing with any "values-aligned" and "security-conscious" projects that come close to building AGI until it has developed it, and to start assisting them to ensure security.
Of course, at the time, OpenAI planned to remain a nonprofit organization. Today, the company is transforming its corporate structure into a more traditional, profit-driven organization. OpenAI reportedly plans to achieve $100 billion in revenue by 2029, equal to the current annual sales of Target and Nestle.
In this context, Altman also said that as OpenAI builds more powerful AI, its goal will be to "lean more towards personal empowerment" while stopping "authoritarian governments from using AI to control their populations through mass surveillance and loss of autonomy ". Altman recently said he thinks OpenAI is on the wrong side of history when it comes to open source technology. While OpenAI has open-sourced some of its technology in the past, the company generally prefers a proprietary, closed-source approach to development.
"AI will permeate all areas of the economy and society, and we expect everything to get smarter," Altman said, "Many of us anticipate the need to empower people to take control of technology more broadly than ever before, including more open-sourcing and accepting that there is a safety vs. personal empowerment the fact that there needs to be a trade-off."
In a footnote to the article, Altman also mentions that OpenAI does not intend to end its relationship with close partner and investor Microsoft by using the term AGI. Microsoft and OpenAI reportedly had a contractual definition of AGI -- an artificial intelligence system capable of generating $100 billion in profits -- and once that criteria was met, OpenAI was able to negotiate more favorable investment terms. However, Altman said OpenAI "fully expects to have a long-term relationship with Microsoft."
