OpenAI CEO proposes the concept of "calculating budget" to allow everyone to enjoy AI dividends equally

In its latest personal blog post, OpenAI CEO Ultraman Sam said that despite the many potentials that artificial intelligence (AI) brings, its benefits may not be widely distributed in the initial stage. He proposed a concept of "computing budgets" that aims to enable everyone around the world to use a large amount of AI technology, thereby ensuring the popularity of technological dividends.

Ultraman notes that historical experience shows that technological advancement often improves many key indicators in the long run, such as health and economic prosperity. However, increasing equality is not the result of a technical decision and may require new ideas to achieve it. He said the balance of power between capital and labor may be disrupted, which requires early action.

He believes that artificial general intelligence (AGI) is coming, and AGI refers to systems that can solve increasingly complex problems at the human level in multiple fields. But Ultraman reminds that these systems may require a lot of human supervision and are unexpectedly poor in some ways.

Although the cost of implementing AGI can be expensive, Ultraman notes that the cost of using the same level of AI will drop about ten times a year. He mentioned that companies like DeepSeek, a Chinese AI startup, showcase the trend of AI models becoming cheaper. He also admitted that huge investments are still needed to achieve AGI and its future development.

Talking about the future of OpenAI, Ultraman said that the company will make some major decisions and restrictions on the security of AGI, which may cause controversy. Previously, OpenAI promised not to compete with any projects close to achieving AGI's "value alignment" and "security awareness", but instead chose to assist them.

Currently, OpenAI is in a transformation, transforming from a non-profit organization to a traditional profit-making organization. Ultraman revealed that OpenAI aims to achieve revenue of US$100 billion in 2029, close to the annual sales of some large companies.

In the process of building stronger AI, OpenAI hopes to gradually enhance personal empowerment while avoiding AI being used by totalitarian governments to control the people. He acknowledged that OpenAI's decisions on open source technology may not be ideal, and he hopes to open up more technology in the future to find a balance between security and personal empowerment.

Ultraman's article comes as the occasion of this week's AI Action Summit in Paris, with many well-known tech figures stating their views on the future of AI.

Blog: https://blog.samaltman.com/three-observations