Nearly every major policy paper, and the wannabe thought leaders that quote them, says that university enrollment and programming skills are the winning combination for the next Industrial Revolution. My analysis of 11 million professional programmers at Gild completely disagreed. This is not the Industrial Revolution. Don’t believe me? Try this thought experiment . . . and be honest.
You are the CEO of a multinational company with 100,000 employees. Rate all of their jobs on a scale from ‘lowest’ to ‘highest’ skill. Now consider a near future in which AI and automation have disrupted the bottom 80% of those jobs by skill-level. Those 80,000 jobs are not needed anymore, and those lower-skilled employees are staring at pink slips. But just as with the Industrial Revolution, automation, in this case in the form of artificial intelligence, has created an equal number of high-skilled jobs. So you have 100,000 employees and 100,000 great jobs—or maybe even more. This is wonderful! Problem solved, right? But wait, now your company needs five times as many high-skill employees. AI hasn’t created any new lower-skill jobs because if they fall below the skills threshold then those jobs are in turn automated as well. So ask yourself these questions: will many, if any, of those lower-skilled employees be qualified to fill these new top-20% roles in your company, even with reskilling?
Take a step back. Today, how easy is it to recruit for and fill those top 20% positions that already exist in your company? How would that change if you have five, ten, twenty times as many “top jobs” to fill? And what if we’re not talking about the top-20% but the top-1%? Will productivity boosts from AI lift your entire labor force into these elite roles? Do you truly believe you can retrain even a minority of your workforce to fill those new jobs?
I believe that we can, but it isn’t going to be through reskilling or the gig economy. It won’t be because we’ve given everyone a university degree or taught them all to program. And in order to secure a robot-proof future for our children and our economy, we must stop pretending that it will be.
Lessons from history
The post-World War II economic transformation in Germany is often cited as the ultimate proof of concept for large-scale reskilling. The successful transition of naval shipyard workers into the booming automotive industry is presented as a template for our own AI-driven disruptions. A closer, more critical look at this historical case study, however, reveals a far more complex and cautionary tale. The success of this grand retooling was highly conditional and exposed a deep, underlying truth about the nature of skills. The retraining programs were overwhelmingly successful for low- to medium-skilled workers whose jobs were defined by relatively routine tasks. For them, it was a lateral transfer; the repetitive work of the factory line was analogous to the repetitive work of the shipyard. They were swapping one set of well-posed problems for another.
The true story lies in the program’s surprising failure. The highly-skilled workers and, most notably, the experienced managers proved profoundly resistant to retraining. This was not a failure of intelligence or work ethic; it was a failure of adaptability. These were individuals with deep expertise in the unique, project-based constraints of building massive vessels. When placed in the high-volume, process-driven world of the automobile factory, their hardwon expertise became a form of cognitive rigidity. They lacked the metalearning skills—the fluid adaptability and comfort with uncertainty—required to navigate a fundamental shift in their professional context. The very brevity of so many six-week retraining programs reveals a systemic misunderstanding of what it truly takes to build these deeper capacities.
The leaders of this transformation, often the scions of the company founders, navigated the chaotic, post-war world with relative ease. They were not just trained in a specific skill; they were raised in an environment that cultivated the very adaptability and strategic thinking the displaced managers lacked, inheriting a form of human capital that prepared them for change. But this post-war boom did not create a universally creative economy. It created a robust, high-skill service economy. This professional middle class was a vital engine of prosperity, but it was distinct from the creative class. This history shows that reskilling for even sophisticated routine work does little to address the persistent, unmet demand for the truly creative talent needed to explore the unknown.
A case study
I saw this exact dynamic play out in a recent collaboration with a major global financial services company. They projected 200,000 layoffs over the next 10 years due to ‘technological obsolescence’ and launched a well-funded corporate initiative to ‘upskill’ their team. The program was a catastrophic failure. Within two years, nearly all of the original one thousand employees had left the company. These elite employees saw the move for what it was: a lateral transition that required an immense amount of effort just to maintain their same professional status. They called it “treading water.” The problem wasn’t a lack of skills. The problem was a fundamental lack of adaptability.
“Reskilling” may be shouted self-servingly as the future of work, but it becomes evident over time that simply reskilling people into different jobs will not improve their long-term prospects because their intellectual experiences have not fundamentally changed. Reskilling, even for the most elite and credentialed workers, is a doomed strategy if their entire career has not prepared them to explore the unknown. Adaptability and other meta-learning skills are not a “soft” layer you can bolt onto an existing skill set in a six-week course. They are foundational capacities that must be cultivated over a lifetime.
But won’t automation free people to be more creative and innovative? Isn’t that what happened in the Industrial Revolution? It’s always struck me that the lazy myth of the Industrial Revolution involves a bit of a bait-and-switch in which new jobs were created and society advanced, and therefore the lives of individual people must have improved at the same pace. Only . . . those weren’t the same people, and at times it took a generation or more for those improvements to take hold. Whenever I hear the AI bait-and-switch it brings to mind a cartoon from, strangely enough, the science journal Nature. Two horses are looking down from a hilltop at a Model-T driving up the road. One horse turns to another and says, “I’m not worried—the wheel, the plow—new innovation always means more jobs for horses.” Unfortunately for our four-legged protagonist, Derek Thompson noted in The Atlantic, “After tractors rolled onto American farms in the early 20th century, the population of horses and mules began to decline steeply, falling nearly 50% by the 1930s and 90% by the 1950s.”
The Gilded Age analogy
The Gilded Age of the late 19th century presents a stark case study in the divergent paths of human capital. The mass migration from farm to factory is often portrayed as a simple story of industrial progress, but it was, for most, a lateral transition into a more brutal form of routine labor. The exhausting, repetitive, and soul-crushing work on the factory line offered regular pay but stripped away the autonomy and seasonal variation of agricultural life, leaving workers with little time or energy for anything beyond sleep and simple diversions. While this new industrial system was insatiable in its hunger for this kind of routine labor, it also created a new landscape of ill-posed engineering and logistical problems, opening a second, very different path for a select few.
The divergence between these two paths was not a matter of luck or circumstance but of endogenous motivation. The individuals who thrived and became the era’s great innovators were not made creative by their new environment; they were spontaneous creatives who brought a pre-existing, fanatical drive to their work. They were the ones already tinkering in the barn after supper, who saw a broken wagon as an opportunity, not a chore. For them, industrialization was a necessary but not sufficient condition for success; it lowered the threshold for their creativity to flourish by freeing them from the necessities of farm labor and exposing them to more complex problems. Their defining characteristic was a willingness to make immense personal sacrifices, forgoing leisure to working endless unpaid hours, not for a specific reward but because they were intrinsically compelled to solve the problem in front of them.
So can we train endogenous motivation and creativity? I’ve been proud of nearly every product that my companies have released, and my work in education most of all. We published scientific papers, gave invited talks, and presented demos around the country with the belief that we would transform teaching. But every teacher that played with our tools said the same thing: “That’s cool . . . and a little terrifying. And what the hell am I supposed to do with it?” We imagined that we were handing teachers a tool to influence the life outcomes of their students.
Each of these innovative products had huge potential to help and could have genuinely been a foundation for a more creative learning experience for both teachers and students. But in every case, they (and I) simply assumed that the presence of the technology would inevitably lead to better outcomes. For all its AI sophistication, cognitive analytics never made teachers or students more creative, even when they were given the freedom to explore without negative consequences. All of these technologies, including my own, were responding to the same basic impulse: because we can imagine a world in which these technologies do good, that world is inevitable.
The myth of technological empowerment
Sadly, it doesn’t work that way. The idea that any technology will make people more creative simply by existing is ludicrous. The vast majority of people, educators included, are heavily entrenched in a pattern of routine labor and systems that discourage creativity. Shoving technology into their hands and saying “go” will not transform work from non-creative into creative overnight.
My research has found intriguing evidence that evoking and developing creativity really is possible, but experiences at Gild and across numerous EdTech projects demonstrated a brutal truth: the idea that technology will magically empower remains pure myth. Of course people can change, but that change comes from intentional effort. It is not the inevitable result of some Econ101 supply and demand curve.
In 2014 at Gild, I had this amazing dataset—122 million working professionals—and my entire mission was to use the data to predict who did the highest quality work. But I’ve never had to interview for a job my whole life. What the hell did I know about what makes a great employee? So when I was hired, I figured I would do what any scientist would do in a job like this—read the existing research. In fact, I read over 100 years worth of research about what makes a great employee. I looked for more than correlations but what actually causes people to do great work. At the same time, I was also the CEO and chief scientist at Socos Learning, where we were looking at the predictors of positive long-term life outcomes in young children.
It turns out, the factors we found in children’s life outcomes were nearly identical to the predictors we found in professionals who did the best work. Perhaps It is not shocking that the qualities that make for an exceptional life also make us good at our jobs. Across the Socos data on children’s long-term life outcomes and Gild’s data on 122 million working professionals, we discovered a rich set of nearly 50 factors that might collectively be described as your ability to learn how to learn. More so than all the data we cram on a resume—your skills, your name, your zip code, even your university—it is these meta-learning factors, things like emotional intelligence, social skills and creativity, that say who you are and who you can become.
Most of these factors involve experiential learning: they develop slowly over time through direct experiences. I know many leaders and venture capitalists in the Tech industry tend to believe that you’ve either got it or you don’t, but research disagrees. Take resilience, defined as an individual’s likelihood of pushing through failure to achieve success. While some of resilience’s qualities are almost certainly rooted in genetic differences between individuals, it is absolutely possible to intervene and increase (or decrease!) resilience over time. But clearly a lecture on the value of resilience won’t change anyone. Instead, a resilience intervention involves direct experience with failure.
And this is where our educational system and labor market get it wrong. Is it important to know when the Treaty of Westphalia was signed or to understand how AI works? Yes, but these are only the tools. We have built our entire education system and labor market not just myopically focused on these tools, but on treating humans as though they were just tools themselves. We educate little kids and employees like they are a tool belt instead of an artist. We hire people this way. This has always been the wrong thing to do. Now we live in a world where AI is a tool that can wield itself. If we continue to build people to be tools in a world where AI is the ultimate tool, we reach a dead end. Our existing institutions do not utilize the strengths of humanity. We need to rebuild education and the future of work to focus on the artist and what they do with the tools. The future of humanity is about the artists.
Excerpted with permission of the publisher John Wiley & Sons, Inc., from Robot-Proof: When Machines Have All the Answers, Build Better People. Copyright © 2026 by Vivienne Ming. This book is available at all bookstores, online booksellers, and from the Wiley web site at www.wiley.com.
