Rodney Brooks, an MIT roboticist and cofounder of iRobot, has sharply criticized the vision of humanoid robot assistants promoted by Elon Musk and other tech leaders, calling it “pure fantasy thinking.” Brooks argued that today’s humanoid robots are fundamentally incapable of becoming the versatile, human-like helpers that investors are betting billions on, as reported by Fortune.
Brooks, who has worked in robotics for more than three decades, says the problem is not a lack of money or computing power, but a misunderstanding of how difficult coordination and dexterity really are. While recent advances in artificial intelligence have transformed speech recognition and image processing, he believes robotics is hitting a wall when it comes to physical interaction with the world.
A key limitation, Brooks explains, is touch. The human hand contains roughly 17,000 mechanoreceptors that detect pressure, vibration, texture, and movement, all processed through multiple families of neurons. This dense sensory network allows humans to perform delicate tasks almost subconsciously. Robots, by contrast, lack anything close to this biological complexity, and there is no large-scale, high-quality dataset for training machines to understand touch in the way AI systems learn from images or language.
Brooks is particularly skeptical of current training approaches used by companies such as Tesla and Figure, which rely heavily on videos of humans performing tasks. The assumption, he says, is that robots can learn dexterity simply by watching. “We do not have a tradition of touch data,” Brooks noted, arguing that visual imitation alone cannot solve the deeper sensory and coordination challenges.
In his view, the massive investment flowing into humanoid robotics is misdirected. He suggested that if even a fraction of that funding were redirected to university research, progress toward useful robots would likely accelerate. Instead, he predicts much of today’s spending will be remembered as an expensive detour.
The criticism comes as Brooks’ former company, iRobot, recently filed for bankruptcy after its valuation collapsed from billions of dollars to a fraction of that level. Despite this, Brooks maintains that practical robots will exist in the future, just not in the form Silicon Valley is currently chasing.
Looking ahead 15 years, he believes successful robots will not closely resemble humans. They may use wheels instead of legs, have multiple arms, or feature specialized grippers rather than human-like hands. They might still be labeled “humanoid,” but their designs will prioritize function over imitation.
As for today’s humanoid robot race, Brooks predicts many of the machines now capturing headlines will quietly disappear. Large sums will have been spent, he says, trying to extract limited performance from designs that were never suited to the task in the first place.