OpenAI CEO Sam Altman has confirmed that the company is working on humanoid robots, with a possible demonstration expected in 2027. Speaking on the Sources podcast, Altman explained why a human-like design could make sense for robots.
“The world is very much designed for people. Matching that form factor seems good,” he said.
This suggests that humanoid robots are better suited to work in environments designed for humans, including homes, offices and factories, where they may need to use doors, tools and other everyday objects.
Altman also hinted that the company could showcase its work soon. “I think we’ll have a cool demo in 2027,” he told Fortune.
However, people may have to wait longer before humanoid robots become a common sight in public spaces.
“I don’t think there will be robots walking the streets for a few years longer,” he added.
Altman did not share details about what the robot would look like or what it could do. However, he said that developing the AI technology that powers the robot was more important than its design.
According to him, the design details were “less important than really figuring out … the brain that makes the robot work.”
Humanoid robots need AI systems that can process information from cameras and sensors, understand their surroundings and decide how to move. These systems help robots perform tasks using their arms, legs and fingers.
OpenAI reportedly established a humanoid robotics lab last year to develop AI models for robots. The company is also advertising 27 robotics-related positions, covering areas such as motion control, whole-robot behaviour, manufacturing and supply chain operations.
These roles suggest that OpenAI is preparing for the challenges involved in developing and producing robots at scale.
The ChatGPT maker will face competition from companies already developing humanoid robots. Tesla has been working on its Optimus robot, while Chinese companies such as Unitree and Xpeng have also made progress.
