“In small factories across America, agile automatons are making everything from parts for AI supercomputers to the hulls of America’s future autonomous naval weapons,” wrote technology columnist Christopher Mims in a recent WSJ article, “America’s Manufacturing Resurgence Will Be Powered by These Robots.” “Once a luxury reserved for big manufacturers, smaller, smarter, more flexible and less expensive ‘cobots’ — collaborative robots — are bringing automation to every fabricator, no matter the size.”
Cobots are part of a broader trend in robotics, — “Specialized robots that use sensors to safely navigate human environments. They can cope with more variability than previous industrial robots, which had no sensing abilities.” Moreover, while programming older industrial robots took years of training, this new generation of industrial robots are radically easier to program, — “now people can use a simple tablet interface to instruct them to perform specific sequences of actions.”
“Manufacturing cobots are a far cry from the humanoid robots which have garnered so much attention of late,” added Mims. “Elon Musk has staked a compensation package that could be worth a trillion dollars on his promise that Tesla will produce a million Optimus robots within a decade.”
“Humanoid robots lack strength, dexterity and specialized appendages — like welders or grinding wheels— but startups building them are betting they compensate through versatility. Perhaps they might unpack boxes in the morning, then pack other boxes in the afternoon. Someday they could, in theory, pick up a welder or a grinding tool, but making them dexterous enough is a huge challenge.”
The appeal of humanoid robots is easy to understand. Since their bodies would be fairly similar to human bodies, they should be able to do everything humans can do given our built-for-human environment, instead of having to develop different special purpose robots for different applications.
But, why is making dexterous humanoid robots a huge challenge, as noted in the WSJ article?
I’ve closely following the evolution of AI over the past few decades, but I have little experience with robotics. So to help me understand why developing humanoid robots is such a challenge I read a few articles on the topic by MIT professor emeritus Rodney Brooks, — who’s long been one of the world’s top robotics experts. Professor Brooks led the MIT AI Lab from 1997 to 2003, and was the founding director of MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) from 2003 until 2007. A robotics entrepreneur, he’s founded of a number of companies, including iRobot, Rethink Robotics, and Robust.AI.
First, I reread “Rodney Brooks’ Three Laws of Robotics,” an essay posted in his website in July of 2024, which I wrote about in this blog.
“Here are some of the things I’ve learned about robotics after working in the field for almost five decades,” he noted:
- The visual appearance of a robot makes a promise about what it can do and how smart it is. It needs to deliver or slightly over deliver on that promise or it will not be accepted.
- When robots and people coexist in the same spaces, the robots must not take away from people’s agency, particularly when the robots are failing, as inevitably they will at times.
- Technologies for robots need 10+ years of steady improvement beyond lab demos of the target tasks to mature to low cost and to have their limitations characterized well enough that they can deliver 99.9% of the time. Every 10 more years gets another 9 in reliability.
His first law — the promise given by the visual appearance of a robot, — tells the buyer or user what to expect. “The point of this first law of robotics is to warn against making a robot appear more than it actually is.” When the robot cannot do all the things its physical appearance suggests, customers will be disappointed. “And disappointed customers are not going to be an advocate for your product/robot, nor be repeat buyers.”
In a second essay posted in May of 2025, “Parallels between Generative AI and Humanoid Robots,” Brooks wrote that the current hype about GenAI and Humanoid Robot may well be related because they’re both based on some very human qualities.
People are able to ask questions of a Large Language Model (LLM) in their natural language on any subject and get back knowledgeable, human-like answers so they assume that the LLM is able to reason like a person. “Similarly, people see a humanoid form robot and its form is an implicit promise that it will ultimately be able to do everything a human can do.”
“It is the apparent human-ness of these two technologies that both lure people in, and then promise human level performance everywhere, even when that level has not yet been demonstrated. People think that surely it is just a matter of time.”
In addition, many now expect that humanoid robots will benefit from technological advances. One of the reasons why deep learning AI, GenAI and LLMs have advanced so much over the past few decades is because the computers needed to train and run AI models have gotten exponentially cheaper and more powerful, thanks to decades-long advances in digital technologies, and more recently, specialized processors like GPUs.
“But you can’t do that with mechanical systems that are doing real work with real payloads,” explained Brooks. “Yes perhaps, just perhaps, today’s physical systems are about four times as large as they need to be to lift and move the objects that a human lifts and moves. Ultimately then we might get a reduction of four times in price,” a far cry from the exponential advances we’ve seen in digital technologies.
“Generative AI and Humanoid Robots tap into the fantasy of infinite wealth from new technologies. Many new technologies have undoubtedly made the lives of humans better, and both these technologies may well do so. But it will not be at the physical scale or short timescale than proponents imagine.”
In a September 2025 essay, Brooks further explained “Why Today’s Humanoids Won’t Learn Dexterity” any time soon despite the efforts and investments of major tech companies. In this long essay, he wrote in great detail about the history and challenges of developing humanoid robots over the past 50 years.
Finally, in the essay’s last section, Brooks made several predictions about how the future of humanoid robots is likely to play out over the next ten to fifteen years, all based on the expectation that “what it means to be a humanoid robot will change over time”:
- Before too long (and we already start to see this) humanoid robots will get wheels for feet, at first two, and later maybe more, with nothing that any longer really resembles human legs in gross form. But they will still be called humanoid robots.
- Then there will be versions which variously have one, two, and three arms. Some of those arms will have five fingered hands, but a lot will have two fingered parallel jaw grippers. Some may have suction cups. But they will still be called humanoid robots.
- Then there will be versions which have a lot of sensors that are not passive cameras, and so they will have eyes that see with active light, or in non-human frequency ranges, and they may have eyes in their hands, and even eyes looking down from near their crotch to see the ground so that they can locomote better over uneven surfaces. But they will still be called humanoid robots.
- There will be many, many robots with different forms for different specialized jobs that humans can do. But they will all still be called humanoid robots.
- And a lot of money will have disappeared, spent on trying to squeeze performance, any performance, from today’s humanoid robots. But those robots will be long gone and mostly conveniently forgotten.
“There are some exciting and promising experiments going on in academic laboratories, but they have not yet gotten close to demonstrating any real dexterity.” But, he reminded us that, as per his third law of robotics, humanoid robots will still need over 10 years of steady improvements beyond lab demos to achieve reliable and profitable deployment even with minimal dexterity.