OpenAI chief outlines three possible endgames for AI development

OpenAI chief executive Sam Altman has outlined three possible paths for the future of advanced artificial intelligence (AI), including one in which a single company becomes powerful enough to dominate its rivals, according to a report by The Information.

OpenAI chief executive Sam Altman has outlined three possible paths for the future of advanced artificial intelligence (AI), including one in which a single company becomes powerful enough to dominate its rivals, according to a report by The Information.

Altman discussed the scenarios with OpenAI employees last week, as governments and technology companies grapple with how to ensure the safe development of increasingly powerful AI systems, the report said.

The first scenario, which Altman appeared to favour, would establish an international body modelled on the International Atomic Energy Agency (IAEA), the United Nations’ nuclear watchdog. Such an organisation would enforce safety standards for leading AI developers and independently assess their systems, according to the report.

The second option, dubbed the “Demis plan” by Altman, refers to an idea associated with Google DeepMind chief executive Demis Hassabis: creating an organisation similar to CERN, the international particle physics research centre. Under Altman’s version, major AI developers would merge into, or work through, a single organisation that would provide AI models to participating countries.

The third scenario, which Altman called the “Eliezer plan”, is named after AI safety theorist Eliezer Yudkowsky. It envisages a company developing AI capable of recursively improving itself and using its resulting economic power to prevent competitors from accessing computing resources.

Altman described the prospect of one company gaining such dominance as undesirable. “Obviously I think that one’s terrible,” he said, according to the report.

The three scenarios come amid growing debate over how to govern advanced AI and manage potential risks.

In the United States, a White House task force known as the Super Intelligence Force, led by Director of National Intelligence Jay Clayton, has begun work on assessing the risks and opportunities posed by advanced AI systems. Reuters reported that the task force held its inaugural meeting in October and established initial protocols and a risk study.