By Tom Carter

Tesla CEO Elon Musk has said the company's Optimus robot will be the biggest product ever. CFOTO/Future Publishing via Getty Images

Jan 22, 2026, 8:29 PM IST

Share

SaveSaved

• Humanoid robots are riding a wave of hype, but the technology is still unproven.
• Three robotics experts broke down how robots can move beyond flashy demos to real-world returns.
• One said they require better sensors and AI models, while another warned about robotics "dirty little secret."

Humanoid robots can do kung-fu and parkour. But can they make your morning coffee?

At a Davos panel on Thursday, moderated by Business Insider's Jamie Heller, three robotics experts said humanoid robots needed to move beyond flashy demos and perform useful tasks in the real world at scale.

Jake Loosararian, the CEO of infrastructure startup Gecko Robotics, said that deploying robots into real-world environments was the major challenge facing the much-hyped industry.

"Deployment is the big problem right now for robotics, in terms of the ability for it to begin to make really large impacts, and for there to be a clear road map," said Loosararian.

The cofounder of the Pittsburgh-based company said that building reliable sources of data about the environments in which humanoid robots were operating would be critical for startups looking to move beyond flashy demos to real-world returns.

Inside Businessnew

Unpacking the biggest stories in business.

Sign up

"You have to forward deploy and build your robots as close to the environment as possible. And that gives you the information and dataset that doesn't exist anywhere on the internet, anywhere on YouTube, about what these environments are like," Loosararian added.

Daniela Rus, the director of MIT's Computer Science and Artificial Intelligence Laboratory, said there was a gap between robots' ability to perform tasks in a laboratory and their ability to take on jobs traditionally done by humans in the real world.

Industry leaders discussed robotics at a Davos panel hosted by Business Insider's Jamie Heller. World Economic Forum

"I can give you a robot that will fold your laundry and load your dishwasher, but it might cost you half a billion dollars," she said.

Rus said bridging that gap would require advancements in a humanoid robot's ability to perceive and manipulate the world through improved sensors and new AI models that can handle situations the robot may not have encountered before.

Shao Tianlan, the CEO of Chinese AI and robotics firm Mech-Mind, said that one big barrier to making humanoid robots useful in a factory environment was developing the capacity for them to learn directly from their human coworkers.

"If we want to deploy a robot, I would say demonstration is the most intuitive way to tell robots what to do. That's exactly how we humans teach others," he said.

Tianlan, who said that Mech-Mind had delivered more than 10,000 "intelligent robots" over the last 12 months, added that he did not believe humanoid robots needed "Einstein-level" intelligence to perform useful tasks, however.

He predicted that robots would take on some jobs performed by humans in controlled environments, such as in logistics and some service industries, in the next "few hundred days."

Humanoid robots are riding a wave of hype as companies like Tesla and Figure prepare to begin large-scale manufacturing of their machines. Tesla CEO Elon Musk has made bold predictions about the company's Optimus robot, saying it could eliminate poverty and be the biggest product of all time.

However, most humanoid robots have not yet been deployed in real-world environments. Many humanoid robot demos are staged in tightly controlled settings, and some use teleoperation, where human operators pilot the robots remotely.

Loosararian called teleoperation the "dirty little secret" of robotics, adding that robotics companies had a responsibility to inform customers about whether their humanoid robots are operating autonomously or with human assistance.

"There's a lot of autonomy for certain tasks, but for the majority of the cases, for humanoids, it's learning in the environment, and it has to do that with teleoperation," Loosararian said.