- Published on
AI breakthroughs in robotics won’t change your life any time soon
- Authors
- Name
- DeepHorus
Humanoid robots powered by artificial intelligence are being promoted as the next transformative technology, but many robotics researchers say the hype is far ahead of reality. Tesla’s Optimus, a white-and-black humanoid, has appeared in videos dancing, passing popcorn, and performing simple chores—while also stumbling, falling, and struggling with tasks like ironing. Elon Musk, Tesla’s CEO, has called Optimus potentially “the biggest product ever” and predicted public sales by the end of 2027 at as little as $20,000 each. In July he told shareholders the robot would eventually have “human and then superhuman dexterity” and could automate almost all human labor, from hauling sheet metal to folding laundry.
Musk is not alone in his optimism. Marc Andreessen, cofounder of Andreessen Horowitz, has said robotics could become the “biggest industry in the history of the planet.” Nvidia CEO Jensen Huang claimed in January that humanoid robots would match human-level ability this year. Morgan Stanley estimates nearly 1 billion humanlike robots by 2050, creating a market worth more than $5 trillion. Such forecasts have attracted intense investment and public attention.
The excitement rests on the belief that AI techniques behind ChatGPT and Claude can give robots human-like movement just as chatbots imitate language. But skeptics argue that language and image intelligence cannot easily master the physical world’s infinite variability. Yann LeCun said at Davos in January that none of the companies building humanoid robots “has any idea how to make those robots smart enough to be useful.” His comment underscored a deep divide between AI evangelists and robotics practitioners.
Researchers also warn against conflating humanoid appearance with general-purpose capability. Jonathan Hurst of Agility Robotics and Oregon State University noted that making a robot look like a person is easy, but making it move and behave dynamically like one is “dramatically more difficult.” The tendency to treat humanoid robots as machines that can learn and perform multiple tasks is misleading, he and others argue. The tension between near-term promises and actual progress is playing out in robotics labs across the United States.
A decade of generative AI breakthroughs has convinced many roboticists that a similar revolution is possible in machines, giving them physical intuition and fluidity. Yet disagreement remains over whether current AI methods are sufficient or whether an entirely new path is needed. For now, the painstaking work in labs suggests that robots capable of navigating the world as humans do are not imminent. The source material is a collaboration between MIT Technology Review and Aventine, a non-profit research foundation.
As companies and investors pour money into humanoid robotics, the gap between promotional videos and reliable real-world performance will shape expectations. Whether Optimus or its rivals reach homes and factories on Musk’s timeline depends less on spectacle than on solving the hard, unglamorous problems of physical intelligence. The question is whether the same techniques that fueled AI’s recent progress will be enough to get there, or if an entirely new path is required.
