OpenAI is one of the frontrunners in the quest to achieve human-level artificial intelligence, and it is reportedly just four steps away from reaching this milestone.
The company has devised a five-level system to monitor its progress towards artificial general intelligence (AGI), which it shared with employees this week, according to an OpenAI spokesperson. The levels range from the current state of conversational AI to AI capable of performing the same volume of work as an organization. OpenAI plans to share these levels with investors and external parties.
OpenAI executives believe they are currently at the first level but are nearing level two, known as “Reasoners.” This level involves AI that can handle basic problem-solving, equating to the capabilities of a human with a doctorate degree but without access to tools. The third level, termed “Agents,” represents AI that can perform various tasks on behalf of a user for several days. The fourth level, “Innovators,” describes AI that can contribute to the development of new inventions.
The company also showcased a research project using GPT-4, indicating it has human-like reasoning skills, as reported by Bloomberg, citing an anonymous source. OpenAI declined to provide further comment.
This system was created by OpenAI executives and leaders who may adjust the levels based on feedback from employees, investors, and the board.
In May, OpenAI dissolved its “Superalignment” team, which focused on addressing the existential risks posed by AI. The responsibilities of this team have been integrated into other research initiatives within the company.