Recent improvements in the dexterity of humanoid robots and of the AIs that control them have led to a surge of publicity about their prospects, not just for industrial uses but as possible helpers in the home. Elon Musk, never one to shy away from a bold prediction, says there will be more humanoid robots than people on Earth by 2040.
So my ears perked up when I heard that a half-marathon in Beijing over the weekend would include a classification for two-legged robots, as a way of showing off all the progress Chinese scientists have made. The robots didn't do so well.
Of the 21 robots that were entered, one veered into a barrier right at the start and shattered, while throwing its human handler to the ground. Another's head fell off. Still another had smoke pour from its head, while one ran in the wrong direction at times, then sat down and declined to get up.
All the robots took large amounts of human attention: changing batteries, spraying water on the robots to reduce overheating, etc. Many had to be tethered to controls held by a human, who ran (or, more often, walked) behind the robots.
Only four of the robots finished in the allotted time of less than four hours, and the fastest took more than 2 1/2 times as long as the human winner, who clocked in at an hour and two minutes.
None of that is to say that humanoid robots have no future. Enthusiasts liken the race to the Grand Challenge for autonomous vehicles held in 2001 that also produced embarrassing results but, 24 years later, has Waymo providing 200,000 fully driverless, paid rides each week in its robotaxis.
But the race does suggest the need for a sober look at the hype about robots, to set expectations for the insurance industry over the next five to 10 years.
If you want to see for yourself what the race looked like, here is a video summary. (The broadcast cuts away after 50 seconds.)
For me, the upshot of the race, in keeping with other progress reports on AI and robotics, is that, no, humanoid robots won't outnumber humans in 15 years. Not even close.
They will be especially scarce in homes, where they will accomplish little while costing as much as a car. (Musk says his Optimus robots will cost $20,000 to $30,000 when they become available next year -- and he has a long history of overpromising.) I dislike doing dishes and laundry, vacuuming and dusting as much as the next person, but I'm not going to pay tens of thousands of dollars to avoid minor chores, especially when my Oura ring keeps telling me to get up and stretch my legs. And you want me to maintain the thing? The extent of my trouble-shooting consists of turning a device off and then turning it on again.
Robots have much better prospects in manufacturing, where they are already a force and are helping workers' compensation carriers and employers keep reducing injuries and, thus, premiums. The robots don't look at all human, but they have automated an awful lot of the assembly in electronics factories and others. Amazon and others use robots to handle much of the grinding work in warehouses.
Progress in manufacturing will continue, likely rapidly, because robots can benefit from improvements in AI while operating in a controlled environment, not having to worry about maneuvering in a small kitchen full of little kids and a puppy.
Even in manufacturing, though, there are limitations. The Wall Street Journal reports, for instance, on how hard it's been for shoemakers like Nike to move work to automated factories in the U.S. and out of Vietnam and China. It turns out that the soft materials in the upper parts of shoes change consistency based on heat and humidity. Skilled human workers can adapt, but robots have trouble. Robots also struggle with the fact that no sole of a shoe is quite the same as any other. They have trouble, too, with the constant changes in shoe design; robots function best when they can finetune their handling of a task and then do it over and over and over and over.
Those of us of a certain age long for Rosie, the maid in "The Jetsons." My daughters tell me the updated version is "Smart House," a Disney movie in which a boy tries to keep his father from dating by programming a house to be a surrogate mother. Or there's "Cassandra," a recent series about a family that moves into a decades-old smart home and reactivates its dormant AI assistant, who was once a human and who was transferred into an AI system.
Whatever your hopes are for robotics in the long term, as you think about the prospects for the next five or 10 years, especially in the home, it's worth keeping in mind this image of the robot that crashed, shattered and threw its handler to the ground only a few feet past the starting line of the Beijing half-marathon:
Cheers,
Paul