Peak robot?

Given our current obsession with the possibility of an economic or even existential robot apocalypse, the news this week that Google is backing away from its aggressive robotics program has received surprisingly little attention. I’m wondering if the company’s retreat might be a signal that, for the moment, we’ve hit peak robot.

Google, according to press reports, is eagerly seeking a buyer for Boston Dynamics, the most vaunted of the robotics companies that it purchased in a wild buying spree a couple years back. Boston Dynamics is famous for making telegenic humanoid and animaloid robots. I’m not sure what practical use the creatures have been put to, but on YouTube they’re superstars:

There’s a certain S&M quality to the Boston Dynamics videos that makes them particularly compelling. Hitting ambulatory robots with hockey sticks and long poles appears to be deeply satisfying.

A strain of sadomasochism also seems to have run through the relationship between the West Coast Googlers and the East Coast Boston Dynamics crew. “The ethos they have and the ethos we have weren’t super-compatible,” Astro Teller, the head of Google’s X lab, told the Wall Street Journal. “They are some of the most talented roboticists in the world, but in order to be here … you have to sign up for our way of doing things.” Ouch.

The problem, though, seems to go a lot deeper than a clash of personalities. Google’s parent company, Alphabet, has dissolved its standalone robotics division, called Replicant, and moved its robotics engineers into X in hopes of “defining some specific real-world problems in which robotics could help,” according to a company spokesperson. That’s hardly a ringing long-term endorsement. The company’s enthusiasm over the practical applications of robots, particularly those with legs, appears to be much diminished.

It may be that we’re about to enter a robot winter, similar to the AI winters of the past, in which a bubble of optimism about technological progress bursts, leaving everyone disenchanted and grumpy. Progress slows until some new breakthrough ignites a burst of new interest and innovation, and sunniness returns. “The core issue we are dealing with here is the realization that making robots that actually do things in the real world is much more difficult than what we had envisioned,” the distinguished French roboticist Jean-Christophe Baillie told IEEE Spectrum. “I tend to believe that we cannot brute force our way to solve the complex problems of interaction with the environment or, even more difficult, with people.”

I’m not saying robots are dead meat. I am saying that an adjustment in expectations may be in order, particularly when it comes to robots operating autonomously or semiautonomously in the real world. Progress in many areas of robotics will probably go slower than we’ve been led to believe. I really hope, though, that Boston Dynamics doesn’t go under. Those videos are great.