[Guest post by Nathan Mizrachi]
KNOXVILLE, TENNESSEE — Inside a giant warehouse that’s bigger than 4 football fields put together, humans unpack boxes from containers before shuttling them onto a series of robotized conveyor belts. The low rhythmic grunts of people working alongside the hum of machines fill this massive space. It’s just another day at Red Stag Fulfilment, a logistics company that specializes in ecommerce fulfillment and employs around 60 full-time and contract workers who inventory and ship everything imaginable to locations around the USA and worldwide.
In 2027, the idea of humans working alongside intelligent machines might seem like a quaint, even backwards notion. Everything from Tesla’s autopilot to Amazon’s Alexa shows that artificial intelligence (AI) is smarter than ever and improving fast. Think pieces from left and right have decried (or celebrated) the inevitable rise of AI. People like Elon Musk have said that yes, eventually the robots really will take our jobs, and it’s the responsibility of society to take care of the newly-unemployed through programs such as Universal Basic Income (UBI).
With the white noise of anti-globalization events such as Brexit and the Trump presidency, we’ve already seen a violent backlash from the formerly steadily-employed masses of the working class. As businesses like Red Stag reach a point where it may become more profitable to deploy a robot or software system to replace human workers, what will the impact be on society and political policy? Then there’s the possibility we’ve all been watching too much Black Mirror, and we’re scaring ourselves into thinking AI is more powerful than it actually is.
“Robots replacing human labor is a sexier idea than the reality that AI software will help humans.”
Jake Rheude is Director of Marketing at Red Stag Fulfillment, and he’s written some really interesting pieces on the future of distribution and how his industry is changing — for the better, he believes. If you don’t know much about how warehouses work, it’s easy to imagine that this is the kind of environment, with its 12m (40ft) ceilings and endless chain of boxes to be unloaded and loaded, where robot technology will be employed. Amazon has released videos of robots moving entire shelves, and images of robot-powered carts that assist human pickers tend to dominate the imagination.
“Robots replacing human labor is a sexier idea than the reality that software will help humans get better at their jobs, rather than taking their jobs outright,” he explains to me over Skype.
At Red Stag, they have 12 different box sizes for employees to choose from when they pack merchandise. The difference between choosing the right size or one that’s too large can mean losing anywhere from 10-30% of their margins on shipping cost based on the incorrect dimensions. The company deploys AI software that calculates the optimal dimensions based on what’s going into a single box.
“Normally an employee needs to have at least 6 months’ experience picking and packing for us before they have that level of familiarity with our operation. We prefer to not invest capital in physical machinery such as robots because there’s a good chance the technology will be obsolete within 3-5 years, whereas AI software can save us much more on our margins while allowing us to cultivate an incentive-based work environment for our employees.” That approach seems to have paid off for this particular company; with steady growth since they were founded in 2013. He concedes that the same model isn’t followed by giants of the industry such as Amazon or Walmart, however. “Amazon might replace their entire workforce with robots if it means squeezing more profitability out of the company.” The same goes for the entire transportation industry, as Uber slowly unveils a driverless fleet and eventually taxi drivers and long-haul truck drivers find themselves replaced by robotized vehicles.
White collar workers and professionals probably don’t need to panic.
It’s inevitable that employment opportunities in the field of logistics will fall, leaving former truck drivers or warehouse packers unemployed. It isn’t a problem isolated to blue-collar fields such as logistics or manufacturing. Eran Abramson, an expert on AI software who works for Israeli startup Knowmail, thinks that surgeons, doctors, bookkeepers and law clerks could all see their jobs phased out by robots or software.
It’s the repetitive motion — the kind of task which can be easily imitated — that’s most at risk in the next 10-15 years of being replaced by software or robots.
Right now, AI technology seems to fall short in terms of improvisation, insight, and creativity. The diagnostic abilities of a talented gastroenterologist are probably more easy to imitate by AI software than the insight into the human personality which guides a behavioral counselor in her sessions with patients. A bookkeeper whose job mostly revolves around using computerized spreadsheets could probably be replaced by a smarter, more intuitive version of QuickBooks software. In the worst case scenario where AI threatens these more skilled fields, Abramson, has some good advice to remember: “What can never be replaced is ingenuity, creativity, empathy, and insight.”
IBM’s Watson has been touted as a possible replacement in the near term for white-collar positions that require strong analytical skills yet fall within a limited boundary of outcomes. A Japanese insurance firm recently replaced 34 claim workers with Watson, and this sets a precedent for similar jobs to be outsourced to software in the future. Does this mean that accountants reading this should be brushing up on their coding skills — another area that could be done by AI — or otherwise diversifying their skillsets?
Don’t panic, says Manuel Ebert. He is the co-founder at machine learning consulting firm summer.ai, and he sees great potential for harmonious collaboration between skilled human workers and AI software. “Most of AI will always be highly specialized and will be wielded by humans who decide where and how to apply them. We have to stop thinking about AI as robots from the Jetsons and Terminator, and realize that AI will look more like Spotify or Roomba.”
For white collar workers, this doesn’t mean you need to worry about Watson taking your job; instead, it’s more likely that Watson or similar AI programs will make your job easier. AI will streamline your workflow will by replying to basic email queries such as scheduling meetings to sync with your calendar. It will cut the time it takes for a specialist doctor to diagnose laboratory tests or MRI scans, and reduce misdiagnoses. It will make the mountains of documents paralegals search through during the discovery phase of casework that much more manageable.
“We’re already cyborgs.”
The time-saving possibilities that AI will bring to many fields is something to celebrate; the less friction between tasks, the better. The ramifications of software that increases efficiency will be far-reaching, touching every sector imaginable. As far as people in tech are concerned, this can only be a good thing. Ebert points out that this isn’t the first time humanity has used glasses to improve vision or “notebooks as external hard-drives for our brains … humans have always been cyborgs who easily fuse with technology.”
If you view artificial intelligence as an enhancement — no different in principle than innovations such as the mould-board plough of Medieval Europe or the feather-light shoes worn by Usain Bolt — then isn’t chapter one. AI has been a protagonist in the ongoing human saga of self-improvement since the first hominids discovered how to make fire. Kentaro Toyama is a professor at the University of Michigan’s School of Information and the author of Geek Heresy: Rescuing Social Change from the Cult of Technology. “Augmented AI in the sense of people using AI to do their jobs is so widespread now that we don’t even think about it. Spell checkers, auto-fill, Google — all have something that could arguably be called AI. It just so happens that we no longer think of them as AI,” he says.
Part of the problem with current pessimistic predictions about AI is they fall into the trap of assuming that, sure, we’ve experienced monumental changes in society before — but nothing else like this. Time and time again, the naysayers are proven wrong. Displaced farm workers move to cities and become part of the manufacturing chain. Increased demand expands the service industry to employ more workers than ever before. Framing the discussion in terms of historical context, and broadening the definition of what we consider AI, quells the panic and doubt. That does provide a bit of comfort to those of us who are worried about the potential millions who could be displaced by machines of one sort of another. But what if this time, our underlying assumptions about how people adapt are flawed? If the mechanisms that were formerly in place have disappeared, will the workers who lose their jobs to automation and AI have the same ability to find jobs like before?
A drop of water in the desert.
There are plenty of experts — including almost everyone I spoke to for this article — who have been optimistic about artificial intelligence. They downplay worries about automation by pointing out that we’ve been here before and that this is just the latest phase of innovation. New jobs will appear, they say, and workers will adapt and learn new skills to keep up with changing demand.
It’s been mostly true in earlier eras of industrialization, as job-seekers migrated to new cities and followed energy, mineral, and industrial booms.
The premise that all workers can be retrained or adapt to a new economy does not seem to apply now as it once may have. Much has been made of how former coal miners in Eastern Kentucky have learned how to code. While it’s encouraging to see laborers adapting to market demand, a trickle of coding jobs to replace a once 26,000-strong sector is like calling a drop of water in the desert a rainstorm. There simply isn’t enough demand for 26,000 coders (or cumulative demand in other sectors). Aneil Tripathy, who studies human behavior and economics at Brandeis University and co-hosts the podcast This Anthro Life, believes that demand isn’t the only factor we should consider.
“Professions such as coal mining, long-haul trucking, or auto manufacturing tend to form a large part of the identity of people employed in those fields. They take a lot of pride in doing what they do; oftentimes it is part of the heritage of their families (“My father and his father before him worked in these same mines that I do today”),” he explains. Contrast this with the upwardly-mobile, usually middle-to-upper-middle class Millennials who might be coders or freelance writers, happy to put in a shift as a barista at their local cafe in between building their portfolio or trying to make it in music journalism.
Unlike a 45 year-old coal miner from West Virginia, whose job is cornerstone of his identity, a 26 year-old living in Brooklyn doesn’t ascribe that same importance to her waitressing gig. To her, it’s something she does to support herself while she works for little or no money as an assistant at an art gallery (plus her parents likely provide a significant monthly stipend to help pay rent). And from a broader perspective, it partly explains why unemployment rates are lower in the cities than they are in rural areas — people who don’t have such a stake in what they do to make a living are probably more able to adapt to a different field.
The kind of jobs added in the aftermath of the Great Recession have been mostly service and hospitality-oriented, with net losses in traditionally solid fields including manufacturing and miniscule gains in mining and logging. Furthermore, the kind of available employment has evolved. Union membership in the USA has declined radically over the last few decades, meaning less working-class jobs are protected by collective bargaining, membership dues, and no-cause firings. Temporary, freelance labor accounts for a larger sum of the labor market than ever before, with a major Princeton study finding that as of 2015, 15.8% of workers were freelancers, contractors, or other non-payroll labor.
With the expansion of gig and sharing economy roles through companies such as Uber, Upwork, Airbnb, and TaskRabbit, a larger portion than ever of employment happens in a way that in the short term has benefited major businesses immensely and contributed to positive growth metrics. Since freelancers don’t qualify for pensions, health insurance, this trend has hollowed out the real disposable income that many wage earners bring home to their households. This transition from union-backed jobs to zero-hour contract labor was documented a few years ago by photographer Ben Roberts, where Amazon opened a warehouse in a former coal mining town in England.
It’s precisely because so many job — both white and blue-collar — lack the safety netting of decades past that the current AI revolution may catch the economy off guard. Not tomorrow, but within a generation it is inevitable that technology will be at the point where our transportation networks will be mostly driverless, manufacturing and logistics will employ even less people than before, and service jobs such as cashiers at fast-food restaurants and groceries will have long been replaced by cheap software programs. The most repetitive of white-collar jobs may follow soon after. Is mere creativity and human insight going to be enough to save millions of people from redundancy? Can the displaced future workers of the world all learn a new skill to such a degree that 100% of them are re-employed elsewhere, or will it be a case of working at Amazon to avoid starvation?
Does a future of discontent await us?
If you combine the scale of net job losses due to automation and the current precariousness of employment for millions of people, you realize we are on the precipice of a man-made yet preventable crisis. “A lot of jobs will disappear quickly due to advances in AI. Note that when this happens, it’s not that everyone will be out of a job, but rather that most people will be out of a job, while others like Elon Musk will become filthy rich,” says Toyama from the University of Michigan.
In the last decade, Western society has already witnessed a clamorous pushback against the status quo. Occupy Wall Street, Podemos, Brexit, and 2016 US Presidential Election are all powerful indicators that a large swath of people are unhappy with the direction our world is going in. The disparity in wealth between the working and middle classes and the richest people at the top, as well as the politicians who are viewed as their lap dogs, has been a major factor in these mass movements.
We should hope for the best and plan for the worst. In the case of artificial intelligence, let’s hope that new technologies are deployed by companies such as Red Stag Fulfillment to the benefit of workers and the broader community. If not, we should expect to see widespread unemployment and even greater discontent from the bottom and middle of society — which threatens the livelihood of those at the top. Now is a good time to think about UBI and what it could look like. Tripathy from Brandeis explains that if UBI is federally-backed and funded similar to the current US Social Security fund, it could stimulate the economy to the same, if not greater, degree that private employment does. “If citizens were offered a guaranteed income, it would encourage investment and spending; the money would feed right back into the economy because recipients would know the next check was coming and spend it, rather than save it.”
A recent article in Quartz interviewed five proponents of UBI from across the political spectrum and their proposals for how it would work vary. A progressive tax on income, plus the elimination of certain welfare programs such as food stamps, could be one solution. There’s also the possibility of funding UBI by levying flat fees on financial transactions or pollution, plus a wealth tax. Given the current political climate in the USA, it’s unlikely that this possible safeguard will be implemented anytime soon.
A society where everyone lives in dignity — regardless of the wages they earn or what they produce — should be the end goal of automation and artificial intelligence, shouldn’t it? “The only reason that we increase productivity and automation is that we can lead more comfortable lives with less fear, insecurity, and inequality. In that sense, UBI has always been the goal of advancing society,” concludes Ebert. Whether the market can meet that goal on its own is a question that will be answered in our near future.