OpenAI has come up with a set of five levels to track its progress toward building artificial intelligence software that can outperform humans, the startup’s latest effort to give people a better understanding of its thinking about the safety and future of AI. The maker of ChatGPT, widely seen as a leader in the race to build more powerful AI systems, shared the new taxonomy with employees on Tuesday during an all-hands meeting, an OpenAI spokesperson said. The levels, which OpenAI plans to share with investors and others outside the company, range from existing AI that can interact in conversational language with people (Level 1) to AI that can perform the work of an organization (Level 5).
According to a spokesperson, OpenAI executives have told employees that the company believes it is currently at the first level, but is on the verge of reaching the second level, which it calls “Reasoners.” This refers to systems that can perform basic problem-solving tasks as well as a human with a PhD but without access to any of the tools.
At the same meeting, company executives demonstrated a research project involving the GPT-4 AI model, which OpenAI says has some new skills that could reach human-like reasoning levels, according to a person familiar with the discussion who asked not to be named because they are not authorized to speak to the press. When contacted for comment on the demonstration, a spokesperson said OpenAI is always testing new capabilities internally, a common practice in the industry. OpenAI has said for years that it is working to build what is commonly known as artificial general intelligence, or AGI—essentially computers that can perform better than humans at most tasks. Such systems do not currently exist. OpenAI CEO Sam Altman has previously said he hopes AGI can be achieved this decade. AI researchers have long debated how to define the criteria for achieving AGI. In a November 2023 paper, researchers at Google DeepMind proposed a framework of five ascending levels of AI, including levels like “expert” and “superhuman.” The ranking is similar to the system often used in the auto industry to rate the level of automation in self-driving cars.
According to the levels that OpenAI has outlined, the third level on the path to AGI will be called “Agents,” which refers to AI systems that can spend days performing actions on behalf of users. Level 4 describes AI that can make new innovations. And the most advanced level will be called “Organizations.” These levels are outlined by OpenAI’s CEO and other senior leaders and are considered a work in progress. The company will gather feedback from its employees, investors, and board, and may adjust the levels over time.
One more thing! We are now on WhatsApp Channel! Follow us there so you never miss any updates from the tech world. To follow HT Tech channel on WhatsApp, click This to join now!
