Are advances in artificial intelligence, robotics and other technologies leading to fewer jobs and more inequality?
This is the first of a two-part special report from People & Power’s Bob Abeshouse. Scroll down for part two.
By Bob Abeshouse
Keep readinglist of 4 items
The United States and other developed countries are in the midst of a digital revolution that may be even more profound than the industrial revolutions of the past. Advances in robotics, cognitive computing and other digital technologies promise untold benefits in a world of leisure hard to imagine. But there is also a dark side to this technological change. It could lead to joblessness for most and extreme inequality, threatening economic health and political stability.
Tension over rising inequality and a lack of good-paying middle class jobs is growing in Silicon Valley and nearby San Francisco, the epicentre of computerisation and the information economy. In San Francisco, buses for Google, Facebook and other companies ferry high-paid tech workers to their jobs in Silicon Valley. This allows tens of thousands to live in the city, fuelling popular anger over gentrification and high housing prices that are pushing longtime residents out.
In San Franciso – as elsewhere in the US – the speed at which top wage earners are pulling away from everyone else is becoming a major issue. According to Martin Ford, author of the new book The Rise of the Robots , “what you see is a few people essentially hoovering up all the income and all the success and everyone else kind of struggling. And one of the implications of digital technologies and the Internet is that more and more of the economy is coming to look like this.”
Ford says advances in digital technology have resulted in job polarisation and “a hollowing out” of middle class jobs. Routine jobs like clerical or assembly line work were the first to be impacted by digital automation, leaving “lower skill jobs that tend to require lots of dexterity and visual perception,” and at the other end “high skill jobs that really require advanced cognitive type skills.” He says that “what we are seeing over time is that both of those poles are gradually being impacted as well. So essentially that hollowed-out middle will get bigger and bigger over time.”
Innovators in Silicon Valley, like robot entrepreneur Steve Cousins, argue that concerns about job loss from technological change are overblown. Cousins, who led one of the most successful robotics incubators in the US, says “there’s not an easy way to stop it and I don’t think it’s desirable to stop progress.” He is now the CEO of Savioke, a company that builds robots for hotels and others in the service industry.
“People have been concerned ever since the Luddites. The technology’s coming, it’s going to take my job,” Cousins says. “And all the evidence historically is that when we have more technology actually the number of jobs grows.”
The Luddites were textile workers in England who rioted in the early 19th century over the introduction of new machines for knitting and spinning. In fact, the industrial revolution gave rise to new businesses and new jobs, raising the standard of living for the masses and producing a middle class never seen before. Tech entrepreneurs and many economists argue that the same will hold true for the digital revolution.
But Ford believes that this time is different. “We have seen major technological disruptions in the past,” he says, “but the issue today is that this technology is … going to invade every sector of the economy. And we’ve never seen machines that can really move into cognitive areas and start competing with people in terms of actual brain power.”
IBM is at the forefront of advances in cognitive computing and machine learning. In 2011, IBM’s Watson computer stunned the world by beating two champions in Jeopardy, a popular TV game show. Since then Watson has become even more powerful, and its capabilities are being marketed for use in a wide range of businesses such as medicine, law and finance.
According to Watson Group Vice President John Gordon, “these are all different professionals that have expertise and we want to think about how we can scale them.” Scaling means breaking down expertise into algorithms so that it can be learned by computers and widely disseminated. Watson has the ability to look through millions of pages of information, “and come up with thousands of different possible options of ways to address the problems and then weigh each of them independently,” Gordon says. “It’s a much broader and more detailed type of analysis than we can do on our own.”
Gordon touts Watson’s machine learning capability. The system’s performance improves with feedback both from humans and the computer itself. “Just like we have more experience as we have more practice,” he says, “these systems can have millions of examples of practice … and improve with experience.” Many are concerned that once humans teach systems like Watson skills needed to perform their jobs that they won’t be needed anymore. But Gordon sees new opportunities for people to collaborate with Watson as leading to enormous growth opportunities.
Watson can also understand and process requests posed in everyday language. Gordon says this is “one of the most powerful aspects of the cognitive computing era. We’re now letting technology understand us as we use natural language and ask questions the way we would ask another person.”
“In the private sector, in the corporate sector, and that’s the area where you really see this drive for efficiency,” he says, “it’s going to mean very often eliminating workers.” Ford also believes that recent advances in cognitive computing that permit computers to write articles and reports as well humans could lead to the widespread loss of white collar jobs.
One of the most sophisticated computer-writing platforms, Wordsmith, was developed by the North Carolina firm Automated Insights. It has a deal to write articles for the Associated Press about company earnings reports, and works with businesses in “everything ranging from business intelligence to finance and sports to health and fitness,” according to company CEO and founder Robbie Allen.
In a blind study, participants could not tell the difference between an article written by the Wordsmith platform, and one written by a human. The company also promotes Wordsmith’s ability to write sophisticated analytic reports for corporate executives, and customised guidance tailored to individual employees.
Allen does not believe his writing platform will have a negative impact on jobs. He admits that the Associated Press would have to hire more people to write the reports Wordsmith does, but says “it’s not cost effective for them to cover 3,000 companies a quarter the good old fashioned way. The only way they can do that is through automation.”
He also argues that “anybody who’s job it is to look at data repetitively and try to make decisions or draw insights from that data, those jobs are not ideally suited for people. And in the future absolutely, I think those types of jobs will be taken over largely by software such as Wordsmith, but I don’t think that’s a bad thing.”
Michael Osborne, a professor of Machine Learning and Artificial Intelligence at Oxford University in England is concerned that technological innovators are not thinking enough about the potential impact of their work on society.
“For those of us working in machine learning,” he says, “I think it’s actually vital that we think about what the societal consequences of our work will be.” Osborne worries that at some point “we will develop algorithms that are at least as competent as human beings in almost all tasks and at that point it’s very difficult to see where the jobs might remain particularly for the low skill workers.”
Osborne and his colleague Carl Benedikt Frey of the Oxford Martin School have studied the potential impact of computerisation on jobs in the US. They conclude that almost half are automatable in the next decade or two.
There is an important difference between the digital revolution and the industrial revolution, according to Frey, that points to potential problems. The industrial revolution replaced skilled artisans with less-skilled factory workers, enabling benefits to be reaped by more people rather than less. Today’s digital revolution mainly provides job opportunities for relatively skilled workers while displacing low and middle skilled jobs. This trend has contributed to the stagnation of wages and “could exacerbate already growing inequality,” Frey says.
Osborne and Frey think their findings are relevant to all advanced economies, not just the US. And they think some of the greatest risk of job loss is in logistics and transportation sector. “Occupations like fork lift drivers, agricultural vehicle drivers, mining vehicle drivers, all these occupations are already being impacted by technologies derived from autonomous vehicles,” Osborne says. These robots are really not far away. In fact, Mercedes-Benz has a self-driving truck it aims to get on the road by 2025. Google rolled out its prototype of an autonomous vehicle last December.
Vivek Wadhwa, a technology expert and fellow at the Stanford Law School, believes that as soon as 10 years from now we will likely be having self-driving cars and robots doing delivery for us. Wadhwa is also a vice president at Singularity University, a think tank and teaching programme funded by successful high tech entrepreneurs and companies like Google, Cisco and Genentech. He argues that digital advances will enable humanity to solve enormous challenges, but that there is also a dark side that most in Silicon Valley are unwilling to face.
“We will have unlimited clean energy .We will have solved the problems of disease and the AI, the artificial intelligence doctors will be better than our doctors are. Good news,” Wadhwa says. “The bad news is that we will need fewer doctors. So we’re now going to create unemployment in the medical industry. The same thing happens with manufacturing. The future I see 15-20 years from now…the vast majority of jobs start disappearing.”
“It’s absolutely no different than when the steam engine came out and people said oh my God, the steam engines going to take away all the jobs,” Cousins counters. “When the car came out and they said there go all the buggy drivers.”
Cousin’s company builds a robot for hotels that delivers toiletries and food to hotel rooms instead of a bellhop. He argues that the robot his company sells “will be a lot cheaper than if you hired a person to do that same thing. But it’s not going to do everything that a person can do. And so what it allows a business to do is add new services.” Cousin envisions his service robots deployed not just in hotels, but hospitals, elder care, restaurants, office buildings.
Wadhwa thinks that the building of robots and other digital advances will have a positive impact on the U.S. economy in the short-term. “I see an economic boom in the United States,” he says. “Then I see jobs disappearing. We’re fine for the next 10, 15, 20 years or so. Beyond 20 years I know we’re in trouble but we have 20 years to figure it out.”