Famous robotics expert Brooks warns: The humanoid robot bubble is doomed to burst.
On September 29th, in the field of humanoid robots, where venture capital is surging and tech giants are vying to enter, renowned robotics expert, co-founder of iRobot, and MIT professor Rodney Brooks recently issued a warning: The current humanoid robot craze, fueled by the hype of companies like Tesla and billions of dollars in investment, is a bubble destined to burst.
01 Three Core Technical Bottlenecks Are Hard to Break: The Gap Between "Pure Fantasy" and Physical Reality
Currently, companies represented by Tesla and the robotics firm Figure are attempting to train robots' fine manipulation abilities by having them watch videos of humans performing tasks. In his latest article, Brooks pointed out that this method is "pure wishful thinking" and dissected its fundamental technical obstacles from three dimensions.
Photo: Renowned robotics expert Rodney Brooks
Firstly, there is a huge gap in tactile perception. Brooks noted that the human hand is an extremely complex structure, densely packed with about 17,000 specialized tactile receptors, which is difficult for any existing robot to match.
Brooks explained that although machine learning has made breakthroughs in speech recognition and image processing, these achievements are based on decades of existing technology that can effectively capture and process relevant data. "But in the field of tactile data, we lack such technological accumulation," he emphasized. This means that robots cannot truly understand the subtle physical feedback generated when humans interact with objects by watching videos.
Secondly, there are unavoidable safety issues. Full - sized bipedal humanoid robots need to consume a huge amount of energy to maintain balance, and once they lose stability, it will pose serious safety hazards. Brooks cited the laws of physics to illustrate that a robot twice the size of the existing model will release eight times more destructive energy when it falls. This potential risk poses a severe test to the large - scale application of humanoid robots in real - world scenarios.
Finally, fundamental problems such as battery life and adaptability to the real environment remain unresolved. Brooks believes that although robots can complete tasks such as folding clothes and warehouse navigation in demonstrations, there is still a long way to go to achieve reliable and cost - controllable large - scale applications.
Brooks' views have resonated in the industry. Some robot enthusiasts admitted on social platforms that the technology "may still take two or three decades to become truly practical," a situation similar to the 3D printing bubble around 2010.
02 A Sober Reflection Amidst Investment Frenzy: The Gap Between Market Expectations and Real - World Implementation
Although the core technical bottlenecks have not been broken through, the capital market's enthusiasm for humanoid robots continues to soar. Figure received capital commitments of over $10 billion in its latest financing round, and its valuation soared to $39 billion.
A market analysis report by MarketsandMarkets predicts that, thanks to the progress of artificial intelligence and the labor shortage in the manufacturing industry, the global humanoid robot market size will surge from $2.03 billion in 2024 to $13.25 billion in 2029.
Investment banks have also given optimistic forecasts. Morgan Stanley expects the market size to reach $5 trillion by 2050, while Bank of America believes that the cost of each robot will drop to $17,000 by 2030.
However, this optimistic expectation stands in sharp contrast to Brooks' sober observation. He pointed out that the current billions of dollars in investment are actually funding expensive training experiments that will never achieve large - scale mass production. He also compared it to "a race where the finish line keeps moving backward, similar to the over - promises of autonomous driving back then."
03 Re - evaluating AI Capabilities from "Generative AI" to Embodied Intelligence
Brooks' skepticism is not limited to robot hardware but also extends to the currently popular field of artificial intelligence. He previously pointed out that the actual capabilities of generative AI are overestimated, and in some scenarios, it may even increase the workload of humans.
This view is supported by empirical research. Experiments by the non - profit AI research institution METR found that software developers using AI tools took 19% longer to complete tasks, but their subjective perception of efficiency increased by 20%, revealing the gap between the efficiency of AI tools in practical applications and users' perception.
Brooks emphasized that applying immature AI technology to the field of robots that require precise physical interaction will further magnify the problems. In a recent discussion on embodied AI, he found that the fundamental problem of "contact manipulation" continues to surface, and how to enable robots to achieve reliable force control and physical interaction in a dynamic environment remains an unsolved issue.
04 The Ecological Dilemma and a Pragmatic Solution: The Wisdom of Survival Under the Shadow of Giants
Although Brooks once questioned the dominance of tech giants in the robotics field, the reality is that leading companies cannot escape their "gravitational pull": Apptronik, a humanoid robot manufacturer, has received investment from Google and is collaborating with DeepMind; Figure has received support from Microsoft and the OpenAI startup fund and has jointly developed an end - to - end AI system.
Facing an ecosystem surrounded by giants, Brooks' prudence stems from his ups - and - downs experience in product implementation. As the creator of Roomba, he witnessed the birth of the most successful household robot in history and remembers well the numerous failures on the road to success. More than 99% of robot projects ultimately failed.
Brooks said bluntly: "In short, developing hardware is really difficult. There are very few executives in humanoid robot companies with actual product deployment experience. I know how difficult it is to deploy robots and how difficult it is to manufacture products that customers are willing to pay for. Products must achieve an extremely high reliability of 99.999%. Any failure rate higher than this will lead to user frustration and even safety hazards."
Therefore, Brooks' new company, Robust.AI, has chosen a more pragmatic approach. The Carter robot of this company abandons the obsession with human form and returns to the essence of an intelligent forklift: it supports autonomous navigation and also allows manual intervention, always placing humans at the core of the decision - making loop.
05 Choices Ahead: Return to the Business Essence or Face the Bubble Burst
Brooks predicted that in 15 years, successful "humanoid" robots will be equipped with wheels, multiple arms, and specialized sensors and will completely abandon the human form. He believes that chasing technological hotspots can boost stock prices in the short term, but ultimately, only the return on investment is the sole criterion for measuring value.
"Customers never care about the technology itself," Brooks emphasized. "What they want are the actual results of cost reduction and efficiency improvement. Whether technology can achieve higher operational efficiency than traditional methods is the key to decision - making."
Regarding the heated discussion about whether humanoid robots will have a "ChatGPT moment," Brooks' warning has sounded an alarm for the industry: Without fundamental breakthroughs in artificial intelligence learning and hardware, the industry will face painful adjustments. Hype will eventually give way to substantial progress; otherwise, this dream of humanoid robots will follow in the footsteps of past technological illusions.
Brooks' warning does not deny the prospects of robot technology but rather guides the industry to return to the business essence. As he asserted that "humanoid robots will be difficult to replace human jobs in the next few decades," perhaps the future winners will not be the most human - like products but practical tools like Roomba that can accurately solve specific problems.
This article is from "Tencent Technology," translated by Jin Lu, edited by Yanni, and published by 36Kr with permission.