Last week, under the tag “Innovations,” the Washington Post reported that Nuro, a Silicon-Valley robotics startup aiming to develop and sell autonomous delivery vehicles, had “announced plans for a new mission: delivering Domino’s pizzas to customers.” The vehicles, which “resemble a giant pill bug on wheels and can reach 25 mph as they operate on major roadways alongside cars,” would zip—or roll, anyway—around Houston and Phoenix, a sort of prepaid mobile vending machine. I have been to Texas, and the idea that a vehicle can operate on a major (or minor, for that matter) roadway at a top speed of 25 mph seemed fanciful. At no point, however, did the paper question that, or, more importantly, the fundamental underlying technological proposition: that these things are really driving themselves.
And yet just last month, in a story on the smaller, cuter, slower “Kiwibots” that now mosey around the UC Berkley campus delivering “burritos, Big Macs and bubble tea,” the San Francisco Chronicle revealed the essential components of the system:
The Kiwibots do not figure out their own routes. Instead, people in Colombia, the home country of Chavez and his two co-founders, plot “waypoints” for the bots to follow, sending them instructions every five to 10 seconds on where to go.
As with other offshoring arrangements, the labor savings are huge. The Colombia workers, who can each handle up to three robots, make less than $2 an hour, which is above the local minimum wage.
“The company,” the Chronicle went on to note, “bristles at comparisons of this to operating remote-control toy cars.” They prefer their own coinage: “parallel autonomy.” Well, every good ventriloquist must maintain that Dummy is a real person in and of his own self.
Technology firms are notoriously secretive about sharing any actual technical operating information about their systems and products, and tech reporters are notoriously shy about poking too deeply into internal operations, for fear of being frozen out of big new product announcements or, worse, losing the chance to take lucrative communications gigs at the very companies they cover. As a result, many stories, like the Post’s, begin from the assumption that these things simply work, and then proceed to report instead on business plans, roll-outs, and potential social impact.
Yet through the cracks, a picture has emerged of an autonomous vehicle industry space whose extravagant claims beg for deeper technical scrutiny. In 2018, Wired published an article entitled, “Self-Driving Cars Have a Secret Weapon: Remote Control,” which it followed up earlier this year with, “The War to Remotely Control Self-Driving Cars Heats Up.” “Autonomy is a vast space,” wrote transportation reporter Alex Davies, “including ride-hail cars, trucks, shuttles, tractors, mining equipment, sidewalk robots, and more. Surely not every company will have the time, resources, or patience to develop their own remote control system.”
Already a year earlier, Nissan had announced at CES, a massive annual consumer technology conference, that truly driverless cars were years—perhaps decades—away and that it intended to solve the problem by creating, in effect, call centers full of drivers ready to take over at a moment’s notice. The reporting in Wired shows just how thoroughly old and new players have leapt into the game, developing technologies to address the problem of signal lag—what the industry calls latency, which could be disastrous and deadly in a large vehicle traveling at even modest speed. But this technology is complex and expensive to develop in its own right, and it seems increasingly like a few dedicated companies will vie to create common remote-control platforms to be used by multiple autonomous vehicle companies.
Davies is a good reporter, and I don’t mean this as a criticism of his reportage, but Wired‘s capacious use of “autonomous,” common in the industry, is misleading, to say the least. And you will frequently find, as you do in the Post story on Nuro vehicles, that insiders and spokespeople frequently slip in (slip up?) the more precise term: unmanned. Which as we know from our own defense establishment’s term-of-art for military drones—UAVs: unmanned aerial vehicles—means very specifically remote controlled.
It is probably too easy to compare every industry fraud to Theranos, the doomed, almost wholly fictitious medical technology company founded by the now-infamous Elizabeth Holmes. What Enron was to the early aughts, Theranos is to the current decade: an easy metonym for just about any scam, especially if it happens to be a questionable startup in a high-tech field.
But it is also hard to avoid the sense that we are in familiar territory with so-called autonomous vehicles, from insectile robots to actual automobiles: a black box in which, as Holmes herself once said, “a chemistry is performed.” An alchemy? An algorithm.
Thernos’ high-tech proprietary gadget, the mystery machine that could perform thousands of blood tests all by itself and almost instantaneously, never worked. It was 100 percent a sham, barely even a bad prop for a cheap science fiction show. In reality, where it did produce test results, Holmes and her company were sending the samples to regular old remote commercial labs, who’d perform standard tests and return the results, which Theranos would pretend to have produced on their machines. Before the full extent of the fakery was revealed, the company would respond to suspicious on this point by saying, in effect, that they did both: automated testing and lab testing as a verification and safeguard.
You might call it “parallel autonomy.”
Actual self-driving vehicles, operating in real built human environments, have proven a far more difficult technical problem than either the genuine optimists or the greedy hucksters of the AV space imagined, and with the scattered revelations about human remote control, you can see a familiar pattern emerging, where the real powers in the industry, the investors looking for profits, are going to turn to the tried-and-proven financial technology of labor-cost arbitrage. If you can outsource your call centers to a low-wage country, surely you can get rid of expensive long-haul truckers and unionized delivery and public transit drivers in favor of roomfuls of low-wage foreigners.
In Isaac Asimov’s 1958 short story, “The Feeling of Power,” he imagines a distant future humanity engaged in a destructive interstellar war with the inhabitants of the planet Deneb. Humanity has long since lost the ability to perform even the simplest arithmetic. All calculation, and all planning, is performed by computers, which, it is implied, largely design and improve themselves.
But then a humble computer technician reverse-engineers basic mathematical operations from the computers’ programming—more as a hobby and pure intellectual exercise than anything else. He calls it “graphitics.” The generals leading the war effort first see it as a curiosity, but by the end of the story recognize an extraordinary opportunity. A man who could calculate in his head could pilot a missile:
[A] missile with a man or two within, controlling flight by graphitics, would be lighter, more mobile, more intelligent. It would give us a lead that might well mean the margin of victory. Besides which, gentlemen, the exigencies of war compel us to remember one thing. A man is much more dispensable than a computer. Manned missiles could be launched in numbers and under circumstances that no good general would care to undertake as far as computer-directed missiles are concerned . . .
The technician, unable to live with himself and this terrible discovery, immediately commits suicide. His boss, who brought this human technology to the generals in the first, does simple multiplication in his head. “And it was amazing the feeling of power that gave him.”