OpenAI Defines Five Steps From AI to AGI

OpenAI has reportedly developed a way to track its progress toward building artificial general intelligence (AGI), which is AI that can outperform humans.

The company shared a new five-level classification system with employees on Tuesday (July 9) and plans to release it to investors and others outside the company in the future, Bloomberg reported Thursday (July 11), citing an OpenAI spokesperson.

OpenAI believes it is now at Level 1, which designates AI that can interact in a conversational way with people, according to the report.

The company believes it is approaching Level 2 (“Reasoners”), which means systems can solve problems as well as a human with a doctorate-level education, the report said.

OpenAI defines Level 3 (“Agents”) as systems that can spend several days acting on a user’s behalf and Level 4 as AI that can develop innovations, per the report.

The top tier, Level 5 (“Organizations”), refers to AI systems that can do the work of an organization, according to the report.

The classification system is considered a work in progress and may change as OpenAI receives feedback on it, the report said.

Two OpenAI executives — CEO Sam Altman and Chief Technology Officer Mira Murati — said in October that AGI will be reached within the next 10 years.

“We’re big believers that you give people better tools, and they do things that astonish you,” Altman said. “And I think AGI will be the best tool humanity has yet created.”

In March, Nvidia CEO Jensen Huang said that, if AGI is defined as the ability to pass any test put in front of it, it could arrive within five years.

“If I gave an AI … every single test that you can possibly imagine, you make that list of tests and put it in front of the computer science industry, and I’m guessing in five years’ time, we’ll do well on every single one,” Huang said.

Elon Musk suggested that AI could surpass the intelligence of the smartest human beings as soon as 2025 or by 2026.

Musk added that the “total amount of sentient compute” — a concept that may refer to AI thinking and acting independently — will exceed all humans in five years.