Op-Ed: Preparing For Work In The Age Of Automation

Relearning is a major priority for humans and artificial intelligence alike.

From warehouse robots to autonomous cars, advancements in robotics and artificial intelligence enter our news feeds everyday. It seems every company is building some form of AI, real or imagined, into its strategy and products. Spectacular Hollywood narratives envision impending doomsday scenarios in which we encounter the singularity or a superintelligence turned against humanity.

The hype surrounding human productivity and singularity through AI, however, is a distraction from the impact that AI and robotics are already having for workers today. A recent report conducted by the McKinsey Global Institute last November found that about half of the work activities people perform could be automated with the current machine-learning algorithms underpinning AI. The workforce is unprepared and untrained for the AI world. Furthermore, those algorithms underpinning AI are critically flawed. Challenges such as these are contributing to marginalization and inequality for workers.

When it comes to the algorithms themselves, Kate Crawford, co-founder of AI Now, noted at a recent MoMA R&D Salon that if you do a search for CEO in any search engine, “you’ll find a lot of images of white men in suits.”

“Depending on which way the algorithm is blowing that day," Crawford continued, "the first woman that you’ll see is, quite often, CEO Barbie.”

Here's another example of an at-best ill-designed algorithm: When MIT professor Joy Buolamwini was researching facial detection software, she noticed a problem: the software didn't detect her face. The algorithm had not been programmed to recognize a variety of skin tones. The “coded gaze,” as Buolamwini calls it, was biased toward the predominantly white images fed to the algorithm.

Bias in coding is downstream from the systemic bias in our academic, government, and corporate institutions. Depending on who gets to make the rules, AI could sustain or exacerbate these systematic biases. This is certainly a question to pose to the large four or five technology firms that are leading the development race of AI.

AMAZING INDUSTRIES, Home Workspaces (2018). Courtesy the author.

Some attempts to tackle these issues only complicate them, such as workers performing low-cost, marginalized tasks to make AI more intelligent. On Amazon’s Mechanical Turk platform, gig workers perform segmented tasks, such as classifying images to train machine-learning algorithms. The average worker on Mechanical Turk makes around $2 per hour, and is essentially training their machine-learning replacements. Amazon has found a way to train its own algorithms, and offers the crowd-as-a-service to others in what Ayhan Aytes calls an “assembly line of cognitive labor.”

The transportation industry is facing its own ethical dilemmas when considering the impact of AI. Mercedes-Benz is choosing to save the passengers first, not the pedestrians, in its autonomous cars, while Uber Drivers, who already face precarity, will need to find new work in the future now that Uber has invested in 24,000 driverless cars.

As Paul Virilio once said, “the invention of the ship was also the invention of the shipwreck.” Advances in AI and robotics can increase human potential and productivity, but they also have Virilian consequences if they are not designed with prescience. Now is the time to ask tough questions about work in the age of automation.

As AI automates rote work across many fields, in what further ways will we be made redundant? Will we suffer jobs losses and lower wage growth? Could AI innovation hurt large numbers of people? How can AI systems be constructed to avoid unconscious bias and even out the inequalities we see in the world today? What changes need to happen in education and financial policy to ensure the wealth from technological productivity in the future is more evenly distributed?

It may be an even more systemic problem than it first appears. The anxiety in the US around AI and robotics is exacerbated by the trends in globalization, financialization, and de-unionization over the last thirty years. While technology has led to a massive increase in productivity and GDP, the number of jobs created and medium household income in the US has been stagnant since 1974. Harold Meyerson wrote that “the tide has continued to rise, but a growing number of boats have been chained to the bottom.” We’re at a point where most Americans don’t have enough cash to cover a $500 surprise emergency.

The rise in freelance jobs won’t necessarily solve the issue either. These jobs don’t have traditional benefits and the pay is uneven. Algorithms also play an increasing role in the gig economy as work is fragmented into bite-sized tasks and dispersed out to workers.

Courtesy Creative Commons/U.S. Air Force. Image: David Dixon.

"Data-driven algorithms for continuous monitoring of worker performance and reputation enable requesters to pick and choose the workers,” says Joseph G. Davis, professor of Information Services and Systems at University of Sydney, in reference to Amazon's Mechanical Turk. “They also have the unilateral right to reject all or part of the work completed by a worker without payment, which adds to the pressure on workers."

Manufacturing in the US is up due to technological productivity, despite the loss of jobs in this sector. These jobs have not moved overseas as the Trump administration claims; they’ve been automated and they are not coming back. The Kiva bots inside an Amazon fulfillment center mechanize the rote process of carrying packages to the warehouse packing station. Computer vision helps them see down aisles, while machine-learning algorithms help them find the quickest paths through the warehouse while navigating robot traffic. Executives from Amazon don’t believe the robots led to job losses; Amazon added 80,000 warehouse employees (for a total of 125,000) in the United States after ramping up 100,000 robots. The jury is out if these bots enable humans to be productive or replace them.

If predictable and physical work will be eaten by machine-learning, perhaps we’ll see a premium on roles like caregivers or artists. This famous study by Oxford University in 2013 shows that jobs with a higher degree of creativity or caring, such as nursing, are very hard for machine-learning to automate. This might generate higher wages in these professions.

The truth is that no one really knows the full impact AI and robotics will have on the US economy or any economy. Even Robert Solow, Nobel-prize winning economist, admitted that “it’s hard to say.” There are two major schools of thought in the debate. One school is ringing the alarm that the progress of AI and robotics will make human-intelligent work redundant because it’s now taking on cognitive tasks and creating a service economy. The other school of thought claims that technological innovation in the past has increased the demand for work and led to higher wages and this time will be no different.

Some bright news for workers is that leading consulting companies are telling their customers that automating out workers will stall growth. Instead, they’re advising companies to focus on augmenting their workforces with AI and robotics helping workers delegate rote labor to machines and take on new tasks that engage them mentally.

Brett Wallace is a New York-based artist whose practice involves an exploration of the future of work. He recently founded AMAZING INDUSTRIES - an R&D startup that demystifies work and advocates for workers in the digital age. The startup as artwork debuted at SPRING/BREAK Art Show and has been reviewed and mentioned in ARTnews, Artslant, Hyperallergic and Whitehot magazine. AMAZING INDUSTRIES is represented by Silas Von Morisse gallery in New York.

Author: Brett Wallace

Brett WallaceSeptember 3, 2018