ANNIE LOWREY picks up the automation theme in a piece at the New York Times, in which she notes that humans have often been complicit in their own obsolescence:
But even more troubling is the fact that crowdsourcing platforms are hurrying along the automation of more and more of these tasks. Erik Brynjolfsson, a co-author of the popular book “Race Against the Machine,” cites image recognition as one obvious place where humans have helped robots replace them. Crowdworkers can collect pennies for identifying adorable cats in photographs, and then companies take that data and improve software that identifies adorable cats with a marginal cost that approaches zero. “We’re at a real inflection point in terms of artificial intelligence and machine learning,” Brynjolfsson said. “Things are speeding up.”
Indeed, many Turkers are actively helping to put themselves out of jobs. “Yesterday it was spam moderation,” said Panos Ipeirotis, a professor of business at New York University. “And today it’s transcriptions and translation. Once we help computers solve the problem of today, we move on to more challenging tasks. Maybe in 10 years, it’s something we think of as completely out of the range of computers right now. I see it happening, all the time.”
It's a good point, but in a way it undersells the power of distributed knowledge to enable automation. Let's turn again to this week's special report:
Google’s expertise at dealing with huge amounts of data will almost certainly play a key part in its plans. By drawing on the computing power of cloud-based systems, its robots, and others, should be able to do much more than they are currently capable of. The self-driving car demonstrates the idea; it can mesh information on its whereabouts from its sensors with maps of the world held in the cloud, with various programs using the comparison to generate instructions for the cars’ motors, steering systems and so on. Ken Goldberg of the University of California, Berkeley, suggests that a similar use of “cloud robotics”—a term coined by a Google employee, James Kuffner—could make it much easier for robots to recognise objects for what they are and act accordingly.
The cloud already houses libraries and programs that can help computers work out what an image is of, and robots to work out from the shape of an object how to pick it up. The European Union’s “RoboEarth” project imagines a cloud-based system that would contain all sorts of such information in a form that robots could use, and that would let robots learn from each other, both about the world around them and about successful ways of tackling tasks in that world.
We are (or at least I am) used to thinking about robots as relatively autonomous contraptions: reliant on the sensors, memory, and processing power they carry on board. That sort of structural autarky represents a huge functional constraint; building machines that must carry the equipment to detect everything they'll need to detect to do their jobs, process the incoming information, and draw upon an internal database that covers most every eventuality greatly limits the functions that can economically be done by robots. But that constraint no longer binds. The world is now blanketed in sensors, most of which are connected to the internet. The machines of the future will be able to draw on that information (or a lot of it, anyway) and use it to inform themselves about their surroundings. They will be able to talk to, learn from, and add to a wealth of data on What Things Are and What To Do With Them. And if they get stuck, they'll be able to talk to humans. The stray object in the path that might once have immobilised a robot won't immobilise it if the robot can ring up tech support and informed that it's just a stray object, go around dummy.
It's very hard to design a machine that can improvise when confronted by the unfamiliar or reason its way through most difficulties—just as it's rare to find a human who can seamlessly navigate his way across all of America's public roads, large and small, without some sort of guide. But just as any regular joe with access to Wikipedia can do a passable impression of someone with enormous intellectual powers, the extended mind of the cloud could lead to impressive improvements in robot capabilities.