The following is an excerpt from Robot-Proof: Higher Education in the Age of Artificial Intelligence by Northeastern University President Joseph E. Aoun. The book calls for a new model of higher education that better prepares students for a global economy driven by artificial intelligence.
In late 2016, the White House’s National Science and Technology Council’s Committee on Technology released a report titled “Preparing for the Future of Artificial Intelligence.” In its heavily footnoted 58 pages, the report offers policy recommendations for dealing with machines’ imminent capacity to “reach and exceed human performance on more and more tasks.”
The report observes that the implications of an AI-suffused world are enormous — especially for the people who work at jobs that soon will be outsourced to artificially intelligent machines.
Although it predicts that AI ultimately will expand the U.S. economy, it also notes that “because AI has the potential to eliminate or drive down wages of some jobs … AI-driven automation will increase the wage gap between less-educated and more-educated workers, potentially increasing economic inequality.” The report ends with a recommendation for further study of the matter.
Such a study might start by examining how technology is already transforming the workplace, changing the nature of skills even in sectors that traditionally have been insulated from automation. For example, the banking business is now comprised largely not of accounting tables but of complex computer models.
The automated space
According to David Julian, executive vice president at Wells Fargo, one of the largest retail banks in the United States, “We have enormous models that have enormous implications to how we manage our business. We’ve got millions of loans, and some system has to calculate the interest.” At a more sophisticated level, he adds, a computer model may attempt to predict losses in the housing market in 10 years’ time by tallying and analyzing quantities of data that are unfathomable by the human mind.
Indeed, today’s banks increasingly need engineers and data scientists. They need to construct complex computer models and understand their inner mechanics so they can test them.
“We can easily test input, data into a system,” said Julian. “We can test the data that we get back. But it’s hard to test that black box. And there’s so much more reliance on that black box. I’ve had to hire a lot more folks with background skills who understand how to open up the box to see if it’s doing what it’s supposed to be doing.”
Those background skills are similar to the ones that might appear on the résumé of a digital media whiz — “math and technology majors,” Julian said of his recruits.
This shift to a reliance on computer models has banks scrambling for talent. Julian’s risk management team has grown from 550 employees to 950 in three years, and he expects to hire another 100 to 200 per year: “They are a very, very sought after commodity. The geeks are the rock stars in financial institutions right now.”
Other professions that once seemed far from the technological sphere are also reacting to its encroachment. For example, today’s media companies, advertising firms, and marketers have enlisted the power of Big Data and advanced software to maximize page views and reach more human eyeballs.
“The media industry is run by robots,” said Grant Theron, executive vice president of global production and partnerships at advertising and marketing giant Young & Rubicam. “It runs on computers and algorithms and targeting.”
William Manfredi, executive vice president of global talent management at Young & Rubicam, agrees that marketing has essentially been transformed into process of data analytics today. “It’s about how you interpret the data to understand people’s behavior. What’s the insight? What’s the ‘creative’ around that, and then what are the channels?”
Increasingly, these channels are automated. “The expression of the creative idea is slowly being pulled into an automated space,” said Theron. When you tap into your device, the technology knows tremendous amounts of information about you, allowing it to customize what you see instantly.
“It’s more of a science than ever before,” agreed Manfredi. “And the people who can see that process from end to end: They are going to be driving the business in the future.”
Working with systems
In contrast to law and advertising, technical sectors like heavy manufacturing might seem susceptible to automation from top to bottom. When Pete McCabe was vice president of global services organization for GE Transportation, he oversaw the services side of a branch of the venerable multinational that constructs, deploys, and manages heavy transportation machinery.
Much of McCabe’s purview was the stuff of complicated software. For instance, his organization manages the flow of traffic on 800-mile-long stretches of single-track railway, determining when to shunt trains to the side to optimize delivery times.
“We have some very, very sophisticated algorithms to drive up to a 10 percent change in velocity, and improvement in on-time schedules and deliveries,” said McCabe. “The difference of one mile per hour for a big railroad is worth $400 million to $500 million.”
Over the past decades, industrial titans such as GE increasingly have moved their business strategies toward software. Instead of staking their fortunes on the sale of giant trains or jet engines or industrial turbines, they now also generate revenue from monitoring them, diagnosing them, and optimizing their performance.
The shift in emphasis from hardware to software has cut costs, improved efficiencies, and upended the skill sets needed by employees. In the past, McCabe said, “99 percent of your workforce was mechanically trained — a trade-school type education in electronics, or mechanics, or HVAC.”
In contrast, today, he is hiring engineers, data scientists, and software programmers. Nonetheless, McCabe added, the skills needed by his software specialists do not begin and end with a fluency in C++. Domain-specific training — even in high-demand domains like data science — is not enough. According to McCabe, “The problems that are going to change outcomes fundamentally, whether it be in productivity, or healthcare, or wherever else, are going to be systems problems.”
For instance, GE recently set out to address the problem of wind blow-over derailments. On open, blustery stretches of the Great Plains, ferocious gusts of wind occasionally hit boxcars at a 90-degree angle, pummeling them clear off the track. This threatens public and environmental safety, ruins schedules, and wrecks the bottom line.
But McCabe knew that GE’s power business had created a model for predicting which trees might fall over in a storm, helping utility companies position their repair trucks in advance. Thus, despite the lack of anemometers on the desolate flats of Nebraska, his group decided to apply the concept to predicting wind blow-over derailments.
“It’s the kind of crazy thing that no single function could solve, or no single kind of profession,” said McCabe. To oversee systemwide projects like this, McCabe was always looking for the Holy Grail of employees — what he calls “the quarterback.”
“I can find engineers, I can find software guys, and I can find good data scientists,” he said. It is harder to find someone who can draw all the threads together to oversee the team of specialists: “Knowing how they plug, knowing where to push. I’d give my left pinky for 10 more of those guys.”
This sort of holistic thought is valued by all sorts of companies. Indeed, it is what all companies look for in managers.
Working with ideas
More and more, managerial abilities such as cross-functionality are now a requirement for entry-level positions. Some companies, such as Google, even make it a cornerstone of their hiring process.
“When you interview for Google, you don’t interview for a job,” said Steve Vinter, engineering director and site lead at Google’s offices in Cambridge, Massachusetts. Vinter said that Google likes to hire generalists. The interview process gauges candidates’ responses to broad challenges, rather than quizzing them on any specific area of knowledge: “How do you think? How do you analyze problems? How do you develop algorithms? How do you measure the performance of those algorithms?”
Because verbal answers to questions like these reveal only so much, Google’s job application process also tests candidates through collaborative problem-solving activities. One of the upshots of this hiring system is that it measures technical expertise and good reasoning skills — but also candidates’ sense of curiosity, their instinct for innovation, and their knack for working well with others.
For instance, one of the defining features of the Google workday is the scrum” — a daily check-in in which everybody stands around and talks about what they’re going to do during the day. “Some of them,” explained Vinter, “are doing things that appear to be unrelated, but you discover their relation by virtue of telling everyone what you’re going to work on.”
This structured serendipity is reinforced by employees having to report back to the group about their daily progress, building on the positive side of peer pressure while also requiring skill in conceptualizing, synthesizing, and communicating ideas.
“Demo days” are another feature of Google’s operational culture — a show-and-tell for extremely accomplished technologists that taps into employees’ pride and inquisitiveness but also their natural tendencies to evaluate the ideas of others and creatively build on them. These human qualities are a crucial component of Google’s astounding success in bridging the gap between the digital and living worlds.
Going forward, people will still need to know specific bodies of knowledge to be effective in the workplace, but that alone will not be enough when intelligent machines are doing much of the heavy lifting of information. To succeed, tomorrow’s employees will have to demonstrate a higher order of thought.
Critical and systems thinking
The definition of critical thinkingis somewhat fluid, but for the purposes of this book, we can say that it involves analyzing ideas in a skillful way and then applying them in a useful one. To do this well, a person needs to be able to observe, reflect, synthesize, and imagine concepts and information and to communicate the results. In short, critical thinking is the desired end product of much of what we do in education.
Machines are getting better at many of the elements that fall under the umbrella of critical thinking, including observation and communication. But they have not grasped all of them. Thus, when a lawyer mulls a thorny contract dispute and figures out how to position a client for a victory and when a marketer crafts the content of a website that keeps eyes on the screen, they are using cognitive capacities that are exclusively human. Critical thinking will therefore remain a cornerstone of human work in the digital age.
Similarly, systems thinking involves seeing across areas that machines might be able to comprehend individually but that they cannot analyze in an integrated way.
Given their intricacy, conceptualizing systems may seem like a task for which digital minds are better suited than human ones, and we do, indeed, rely on computers to understand complex networks.
But computers cannot decide what to do with that information. For example, a computer can model climate change, but it takes human beings to devise and enact policies to stem it. Likewise, in the case of wind blow-over derailment, computers can help a team of engineers predict when it is likely to occur, but they cannot marshal the different talents needed for the project, give them direction, interpret the wider ramifications of the findings, and decide how to implement change. Indeed, a computer would not have had the idea for the project in the first place. Only humans exhibit that sort of creativity.
Because critical thinking and systems thinking are crucial for the human employees of the future, it is imperative that we instill them through the education of the present. Universities will have to develop methods to nurture these cognitive capacities in students if they hope to maintain their age-old social compact, equip graduates for fulfilling, productive lives, and generate new knowledge. To compete with intelligent, advanced machines, we will need to think intelligently about advancing higher education.