Future Imperfect #27: The rise and (potential) fall of crowdworking
Welcome to Future Imperfect! This week I’ve been following the rise of crowdworking (followed by the fall of crowdworking), quantum computing, modern infantry tactics, and your driving fingerprint.
If this is your first time here, be sure to follow the FI publication to get this newsletter in your inbox each week!
The rise and (potential) fall of crowdworking
From Mark Harris via Backchannel, Crowdworking (or as Amazon’s Mechanical Turk calls it, “human intelligence tasks”) is increasingly taking away a lot of the repetitive jobs that would have otherwise gone to highly-skilled workers.
The engineers who are developing this model of labor have a bold ambition to atomize entire careers into micro-tasks that almost anyone, anywhere in the world, can carry out online. They’re banking on the idea that any technology that can make a complex process 100 times cheaper, as in Harry’s case, will spread like wildfire.
Perhaps it’s inevitable that in a few years, software will swallow up these jobs, too. But as the tech conversation has fixated on how artificial intelligence will affect the job market, crowdwork has quietly grown in impact and scale. The next jobs to receive the crowd treatment? Doctors, managers and teachers.
But lets pause on the software/algorithm aspect. I don’t think it’s credible to claim that this is going to become a new normal in the world economy, or at minimum in the American economy, simply because the actors who are thinking about automation are already at work. Several of the interviewees from the piece call this out as a simple inevitability.
“Algorithms are going to take a piece of the work,” admits Isaac Nichols. “It’s a slow, powerful steady process,” agrees Lukas Biewald. “What we see is that people will move little pieces to automated systems over time.”
Adam Devine, a vice president at WorkFusion, another crowdworking platform, goes further. “There is absolutely no future where a person is reading information and simply keying in data.” For example, one of WorkFusion’s clients processes payment records between banks in different countries. These documents arrive from global banks in emails, spreadsheet files, PDFs and even faxes. Workers then have to transcribe each one perfectly. According to Devine, a single mistake can cost half a billion dollars.
Devine says that by toggling back and forth between a human worker training an algorithm, and an algorithm that makes a mistake and is corrected by a human, WorkFusion can achieve near-perfect accuracy at a fifth of the cost of using people alone. And as machines get smarter, the illegible writing and scrawled numbers that currently need human input will get fewer and fewer. “The workers don’t even realize that they’re working on an interface that’s ultimately going to remove a lot of the work they’re doing,” says Devine.
Why does this matter? In a recent a16z podcast, Kevin Kelly, founding executive editor of Wired, piped in on a similar subject. In his view, “any job that can be specified in terms of efficiency or productivity is a job that humans should not be doing, and would be a job that goes to the bots.” That pretty much sounds like the definition of this sort of crowdworking to me.
Quantum Leap
You’ve probably heard about the quantum computer hype from time to time. It’s time for a little bit of clarification on what these devices will, and will not, be able to accomplish. Stephen Ornes, via Quanta, on computing’s search for quantum questions.
What can these computers do that classical computers can’t? The claim of a 100-million-factor speedup did not conclusively prove that the D-Wave 2X — and quantum annealers in general — will profoundly surpass the abilities of classical machines. A case in point: The paper announcing the results was careful to mention that the 100-million-factor speedup came when the D-Wave computer was pitted against one particular type of algorithm running on a classical computer. Change the algorithm to a more efficient one, and the speedup disappears. “It’s a little like saying, ‘OK, we’re going to have a motorcycle race. Everybody bring out your motorbike.’ But only one person knows it’s going to be on dirt,” said Helmut Katzgraber, a computational physicist at Texas A&M University. “Then they bring the dirt bike, but nobody else knows. That’s basically what’s been done there.”
So how would the D-Wave machine compare in a fair race against the fastest classical computers? It depends on the racetrack.
Computer scientists are now actively mapping out so-called “benchmark problems” — the classes of problems that are particularly suited to the type of hybrid quantum machines epitomized by the D-Wave 2X. A study co-authored by Katzgraber and posted to the scientific preprint site arxiv.org in April concludes that scaled-up quantum annealers should be able to outperform classical computers in certain narrow computing domains. Fortunately, these domains are likely to include important problems in machine learning, protein folding and route planning, to name a few. But exactly which of these problems will show a marked improvement when processed by a quantum annealer, and how fast the speedup will be — these are questions that computer scientists are only beginning to understand.
Why does this matter? Quantum computers aren’t a panacea for getting back to, and surpassing, Moore’s Law because they aren’t necessarily best used as a general purpose computing tool, but that doesn’t mean that they aren’t worth attention and won’t have incredibly useful applications.
The anti-numbers game
Infantry, the most basic building block of military throughout history, is truly entering the 21st century (via War on the Rocks).
Since the firearm’s creation, firepower has aggregated at increasingly lower echelons in armies with each century, allowing smaller number of soldiers to dominate larger amounts of terrain and inflict greater amounts of damage on enemies. Today, a pair of soldiers handling a 20th-century machine gun can exceed rates of fire achievable only by an entire regiment of 19th-century riflemen. Advances in robotics, miniaturization of technology, and most importantly, exploration of human-machine combat teaming concepts will continue this trend and render the infantry squad of the 21st century the deadliest yet.
For the most part, these advances do not bring new capabilities to land forces; they simply miniaturize existing capabilities. Legged and tracked robots provide an avenue to place tank-like capabilities at the squad level. Precision indirect fire and close air support, once the domain of divisions and brigades, can be inexpensively placed in the hands of a platoon leader guiding remotely piloted aircraft. Modern infantrymen have always coordinated with armor, artillery, and airpower, but these new systems under development will be under the organic control of infantrymen themselves, creating an opportunity for even greater synergy in combat arms and greater lethality. These technological advances will trigger the most substantial changes in infantry tactics since World War II.
Why does this matter? The military is not a numbers game anymore. It’s not about who has the most troops on the ground or the most ships at sea (not that politicians care to observe). Maybe it’s still a hangover from our society’s obsession with World War 2 where the numbers maxim still proved true. Either way, the advanced technologies described here, from airborne drones to on-the-ground robots, will further empower the few against the many.
Eyes on the road
Privacy advocates and Maryland drivers beware: A.I. is going to find you based on your terrible driving.
In a study they plan to present at the Privacy Enhancing Technology Symposium in Germany this July, a group of researchers from the University of Washington and the University of California at San Diego found that they could “fingerprint” drivers based only on data they collected from internal computer network of the vehicle their test subjects were driving, what’s known as a car’s CAN bus. In fact, they found that the data collected from a car’s brake pedal alone could let them correctly distinguish the correct driver out of 15 individuals about nine times out of ten, after just 15 minutes of driving. With 90 minutes driving data or monitoring more car components, they could pick out the correct driver fully 100 percent of the time.
Why does this matter? This reminded me of the website that could identify unique individuals on the web through just the active applications and device settings identifiable through public means. If we can be publicly identified by our driving habits, then there really is nowhere left to hide — unless we all make for the woods that is. Though some good may come of this through anti-theft measures, I’m not sure that the overall balance is positive.
That said, we’ll all have autonomous vehicles before long, so maybe it’s a moot point.
Don’t forget to save
From Terraform, a brief check-in with parents who don’t want to let go of their children.
I’m still holding Bessandra in the air, hoping my hands under her armpits are more comfortable for an infant than they would be for a full-size adult, when my sister walks in, also holding Bessandra.
I look between the two identical babies and hope my meds aren’t responsible. “Uh, you had twins?” I say, even though I’m pretty sure she did not have twins — the past four months have been a non-stop Stream of one baby, not two. Covering that up would be a lot of work for a prank.
“Surprise,” Maggi says dryly. “No, you’re holding our SP.”
“Your what?”
“Save Point,” Maggi beams.
I could easily imagine this getting very creepy — kind of glad the author didn’t go there.
GIF of the Week: Outsmarting the real children
Weather Forecast: Heating up with a 24 percent chance of Brexit
Like Future Imperfect? Why don’t you go ahead and click that recommend button below so others can find it too! And send me feedback at lasky.joshua@gmail.com.
