OpenAI CEO Sam Altman announced during a live broadcast on Tuesday that OpenAI's deep learning system is making rapid progress, and the efficiency of the model in solving complex tasks continues to increase. Altman said that the company expects to achieve "intern-level research assistants" by September 2026, and strives to develop fully automated "legal AI researchers" by 2028.
According to reports, this goal refers to the system being able to complete large-scale research projects independently, rather than simply assisting human scientific research.
This announcement comes as OpenAI completes its transformation from a non-profit organization to a public benefit corporation structure (Public Benefit Corporation), breaking the restrictions on fund raising and other aspects of the original non-profit charter, helping to promote the entry of more capital and infrastructure expansion. Jakub Pachocki, the company's chief scientist, further explained in the live broadcast that artificial intelligence in the future is expected to achieve "super intelligence" in less than ten years, that is, its ability to surpass humans in many key areas. He pointed out that the current OpenAI model can already compete with top human players in competitions such as the International Mathematical Olympiad, and its ability to complete a task of about five hours has reached a high level.
In order to achieve these goals, OpenAI is accelerating breakthroughs through continuous algorithm innovation and greatly increasing the "test time compute" (that is, the computing resources and time for model thinking about problems). Pachocki said that important scientific breakthroughs in the future will require the use of entire data center-level computing resources for a single problem, which will greatly push the boundaries of intelligence.
Altman added that this organizational restructuring has laid an institutional foundation for the company to accelerate the advancement of AI research assistants, and pledged to continue its commitment to responsible AI development. Under the new structure, the non-profit OpenAI Foundation, which focuses on scientific development, will hold 26% of the shares of the profit-making entity and lead the research direction. The foundation has received a funding commitment of US$25 billion, focusing on scientific research tasks such as using AI to conquer diseases, and assisting in the development of AI safety and social responsibility plans.
In terms of infrastructure construction, Altman revealed that OpenAI has promised to be equipped with 30 gigawatts of data center computing capacity in the next few years, involving $1.4 trillion in financial expenditures. He bluntly said, "OpenAI hopes to invest up to one trillion US dollars in infrastructure every year in the future" to help promote fundamental progress in the field of science and technology.
OpenAI stated that the plan to create automated AI researchers is in line with the company's overall strategy to promote scientific research, promote AI to surpass human discovery, solve complex problems that are currently difficult for humans to overcome, and accelerate innovation in many industries including medicine, physics, and technology development.
