TECHNOLOGY & ARTIFICIAL INTELLIGENCE — The founder and CEO of Chinese robotics company Unitree says humanoid robots could be several years away from reaching their own “ChatGPT moment” — a breakthrough that would allow robots to understand instructions and perform a wide range of tasks in unfamiliar environments.
Wang Xingxing said the technology could reach that important turning point within roughly two to ten years, highlighting both the rapid progress of robotics and the significant technical challenges that remain.
What Would a “ChatGPT Moment” Mean for Robots?
For humanoid robots, the phrase refers to a major improvement in artificial intelligence that would allow a machine to generalise its knowledge rather than simply perform a limited number of programmed actions.
Instead of requiring detailed instructions for every individual task, a robot could receive a simple voice or text command and work out how to complete it.
The robot would also need to operate effectively in environments it has never encountered before.
The 80 Percent Challenge
Wang has described a future in which a robot could successfully carry out around eighty percent of requested tasks in an unfamiliar environment.
Reaching that level of generalisation would represent a major change from today's robots, many of which perform specific tasks under controlled conditions.
Once robots can reliably understand their surroundings and adapt to new situations, Wang believes commercial deployment could expand rapidly.
Today's Robots Are Impressive but Limited
Humanoid robots have made remarkable progress in movement, balance and coordination.
Machines can now walk, run, dance, perform athletic movements and interact with objects in increasingly sophisticated ways.
But physical ability is only one part of the problem.
A robot that can perform a backflip is not necessarily capable of preparing a meal, cleaning an unfamiliar room or safely handling unpredictable objects without human assistance.
The Real Challenge Is Robot Intelligence
The next major breakthrough is likely to come from the software controlling robots rather than simply from stronger motors or better mechanical designs.
Robots need systems that can combine vision, language, reasoning, movement and knowledge of the physical world.
This field is often described as embodied artificial intelligence, where AI systems learn not only to process information but also to interact with the physical environment.
Unitree Is Betting on World Models
Unitree is investing heavily in technologies known as world models, which are intended to help AI systems understand how objects and environments behave.
Such systems could allow robots to predict what might happen after taking an action and select an appropriate response.
That capability could become essential if humanoid robots are expected to operate independently around people.
China Is Moving Quickly
China has become one of the world's leading markets for humanoid robots and has developed a large ecosystem of robotics manufacturers and suppliers.
Chinese companies are producing robots for research, industrial applications and demonstrations while attempting to reduce manufacturing costs.
The country's manufacturing infrastructure could give domestic robotics companies an advantage if humanoid robots eventually enter mass production.
Unitree's Rise Shows the Scale of the Industry
Unitree has become one of China's most prominent robotics companies, producing both quadruped robots and humanoid machines.
The company recently made a major stock-market debut in Shanghai, attracting enormous investor interest and raising substantial capital for future development.
The market reaction reflects the enormous expectations surrounding humanoid robotics, although today's commercial applications remain far more limited than the excitement surrounding the technology might suggest.
Mass Adoption Is Still a Difficult Problem
The biggest question facing the industry is not whether humanoid robots can be built.
It is whether they can become useful enough and affordable enough for millions of people and businesses to buy them.
Current humanoid robots remain expensive, require significant development and are often used primarily for research, demonstrations and controlled industrial tasks.
That means the industry still has a considerable distance to travel before humanoid robots become ordinary household or workplace equipment. 3
Why the Breakthrough Could Change Everything
If robots eventually acquire reliable general-purpose intelligence, their economic impact could be enormous.
A single machine could potentially perform multiple tasks instead of being designed for one narrow function.
Factories could deploy robots across different production lines, businesses could use them for logistics and service work, and households could eventually use them for a broad range of domestic tasks.
The economic value would come not simply from having humanoid machines, but from giving those machines the flexibility of human workers.
The Difference Between Demonstrations and Real Work
The robotics industry has become famous for spectacular demonstrations.
Robots can dance, run, fight, jump and perform carefully prepared sequences that attract enormous public attention.
But real workplaces are much less predictable.
A commercially useful robot must deal with clutter, unexpected objects, changing instructions, human workers and equipment it has never encountered before.
That is why generalisation may ultimately matter more than athletic performance.
Artificial Intelligence Could Accelerate Robotics
The rapid development of generative AI has changed expectations for robotics.
Large AI models have demonstrated that machines can learn patterns from enormous amounts of data and respond to natural-language instructions.
The robotics industry is now attempting to transfer similar capabilities into physical machines.
The challenge is much harder because robots must understand the physical world and execute actions safely in real time.
The Ten-Year Timeline Is Not a Promise
Wang's estimate should be understood as a range rather than a prediction that a specific breakthrough will definitely occur within a particular year.
Progress in robotics depends on advances in AI, sensors, computing, batteries, motors, training data and manufacturing.
A breakthrough in one area may not be enough if another part of the system remains inadequate.
The technology could advance faster than expected, or important obstacles could push widespread adoption much further into the future.
What Would Prove the Industry Has Reached the Moment?
The clearest evidence would come from robots performing useful work independently in unpredictable environments.
If a robot could enter an unfamiliar workplace, understand verbal instructions, safely manipulate objects and complete most tasks without specialised programming, that would represent a genuine turning point.
At that stage, humanoid robots would begin moving from impressive technology demonstrations toward general-purpose machines.
Our Perspective
The important point is that the robotics revolution will not be decided by which company builds the robot with the most impressive movements.
The winner will likely be the company that solves the much harder problem of intelligence, reliability and cost at the same time.
A humanoid robot that can perform almost any useful task in an unfamiliar environment would be far more important than a robot that can simply walk, dance or run faster.
Conclusion
Unitree founder Wang Xingxing believes humanoid robots could be between two and ten years away from a “ChatGPT moment” in which they become capable of understanding instructions and completing a large share of tasks in unfamiliar environments. 4
The technology is advancing quickly, but major challenges remain in artificial intelligence, physical reasoning, reliability, cost and real-world deployment.
If those challenges are solved, humanoid robots could move from impressive demonstrations to genuinely general-purpose machines, potentially transforming workplaces and homes on a scale similar to the impact of modern AI.
Daily Touch Insights Editorial Team
View Journalist Profile
