The Wrong Question
On AI, layoffs, geopolitics, and the real bottleneck nobody wants to talk about Layers Not Linear · Tyler Blackwell · 10 min read
What does it take to be on the right side of this transition - and are we building the systems that give people a real shot at getting there?
That question is going to matter more in the next ten years than almost any other question a person, an organization, or a government can ask. And right now, most of us are asking a different one entirely.
The question most people are asking is: what jobs will AI destroy?
It is the wrong question. Not because the answer doesn’t matter, but because it directs our attention backward - toward what is being lost - when the more important and more urgent question is forward-looking. What does it take to be on the right side of this? And are we, honestly, building toward that?
I don’t think we are. Here is why.
The number nobody talks about
In March 2026, AI was the leading cause of announced U.S. layoffs, accounting for roughly a quarter of the total - 15,341 jobs in a single month. Ken Griffin, founder of Citadel, said recently that work his firm would normally do with master’s and PhD-level finance professionals - work that used to take weeks or months - is now being done by AI agents in hours or days. “These are not mid-tier white collar jobs,” he said. “These are extraordinarily high-skilled jobs being automated by agentic AI.” He said he went home one Friday depressed about it.
I understand that Friday. I’ve had versions of it myself.
But here is the number that doesn’t make the headline: hundreds of thousands of AI-related jobs are sitting unfilled right now. Positions requiring generative AI skills have quadrupled over the past two years. IDC estimates that AI skills shortages will cost the global economy $5.5 trillion by 2026 in missed revenue, product delays, and impaired competitiveness. Nearly 90% of organizations now use AI in their operations. Only 9% have reached true AI maturity.
Read that again. Nine percent. The tools exist. The demand exists. The jobs exist. The people who can do those jobs do not yet exist in sufficient numbers.
This is not an AI problem. This is an education and training problem. And it has been building for years while we debated the wrong things.
The steam engine analogy is a good one - but not for the reason people think
The tweets comparing this moment to the Industrial Revolution are everywhere and mostly correct. They’re just drawing the wrong lesson.
When the steam engine arrived, it did not simply replace human muscle. It created entirely new categories of work that hadn’t existed before. It reorganized geography - factories could now be built anywhere, not just near rivers. It reorganized time - work could happen continuously, independent of daylight or weather. It reorganized society in ways that played out over generations.
But here is what the steam engine analogy misses when applied to AI: the Industrial Revolution’s disruption was primarily about physical skills and abilities. It replaced hands and backs. The people displaced from agricultural labor could, over time, learn to operate machines. The gap between old skill and new skill, while painful, was crossable.
AI’s disruption is cognitive. It is replacing analytical tasks, pattern recognition, synthesis, and increasingly, judgment. The gap between the old skill and the new skill is not about learning to push a different button. It is about developing a fundamentally different relationship with knowledge work itself - learning to direct, evaluate, and augment AI systems rather than simply perform the tasks those systems now handle.
That is a harder gap to cross. And it requires a different kind of education than we are currently providing.
Which brings us back to the real question: are we building the systems that give people a real shot at getting there?
The honest answer right now is no.
The geopolitics issue not being connected to the workforce
This is where I want to zoom out further than most workforce conversations go.
The AI race is not just a technology competition. It is a geopolitical competition. And the countries that win it will not necessarily be the ones with the most advanced models. They will be the ones with the most prepared workforces.
Areas with higher education levels - measured by literacy, numeracy, and college attainment - are significantly more likely to adopt the new AI-related skills reshaping labor markets. Areas with lower educational attainment are being left behind in the adoption curve. That is not a technology story. That is a geography and inequality story wearing a technology costume.
China, South Korea, Singapore, and the UAE are treating AI literacy as a national security issue, not just a workforce development issue. They are right. The United States, meanwhile, was having a debate about TikTok while the underlying infrastructure question - how do we prepare 160 million workers for an AI-native economy - went (and still is) mostly unaddressed at the policy level.
The geopolitical risk is not that China builds a better model than OpenAI or Anthropic. The geopolitical risk is that other countries build better workforces than ours.
An important aside on this point: China started this race in 2018, when its Ministry of Education released national guidelines integrating AI into the K-12 curriculum - seven years ago. Since September 2025, every student in the country, from first grade through university, receives mandatory AI education as a core subject, tiered by age: interactive games and storytelling in primary school, ethical dilemmas and real-world applications in middle and high school, interdisciplinary AI coursework in college.
The government has developed standardized AI textbooks, partnered with Baidu and Huawei to bring industry into the classroom, connected virtually every school in the country to the internet, and built digital management platforms in more than half of its schools. One former New York state education official who visited Shanghai classrooms in late 2025 said China is approximately a decade ahead of the United States in AI education implementation. The United States, by contrast, has no national curriculum.
Education is governed by more than 13,000 independent school districts, each making its own decisions about spending, hiring, and what gets taught. California was the first state to mandate AI literacy - in 2025. A White House directive requiring K-12 schools to introduce AI education was issued in April 2025, but a mandate without infrastructure, teacher training, or funding is largely symbolic. Eighty-one percent of U.S. teachers say they lack the time to develop an AI training curriculum. Seventy-five percent say they lack the knowledge. The gap is not a difference in ambition. It is a difference in architecture - and in the willingness to treat AI literacy as the national infrastructure investment it actually is.
So, this is an issue of systems. Are we building the systems that give people a real shot? Right now, the education system in the U.S. is the bottleneck. And we are not treating it like one.
The real scandal inside the layoff numbers
Let me complicate the AI-layoff narrative further, because the numbers deserve scrutiny.
Oxford Economics found that many firms are attributing layoffs to AI because it conveys a more positive message to investors than admitting to over-hiring or weak demand. The actual AI-attributed job cuts, while real, represent just 4.5% of total reported job losses. By comparison, job losses attributed to standard market and economic conditions were four times larger.
Meanwhile, 55% of employers already report regretting their AI layoffs. Klarna replaced 700 employees with AI, quality declined, customers revolted, and the company had to rehire humans. Forrester Research predicts that half of all AI-attributed layoffs will end up being quietly reversed - with workers rehired offshore or at significantly lower salaries.
A significant portion of what is being called AI displacement is post-pandemic over-hiring correction dressed up in technology language. Companies attributing layoffs to AI because it makes a bad news story sound like a good news story to their investors.
This does not mean AI displacement isn’t real. It is. Griffin’s experience at Citadel is real. My own experience building agentic workflows and watching tasks disappear is real. But the current moment is messier than the narrative suggests. Organizations are cutting people for AI capabilities they don’t fully have yet, in an economy where the workers who could actually use AI well are simultaneously the hardest to hire.
Only 16% of individual workers had high AI readiness in 2025. Only 23% of AI decision-makers say their organizations offered prompt engineering training. We are cutting the people and not building the replacements.
That is not a transition strategy. That is a bottleneck in disguise.
The entry-level trap
Here is a dimension of this that I think about a lot, and the one that will define the next decade of labor markets more than any other.
The traditional deal of entry-level work - trading rote labor for mentorship and development - is breaking down. AI is eliminating the grunt work that early-career workers used to learn on: code generation, financial modeling, research synthesis. Early-career professionals are stranded between AI agents doing the tasks they were hired for and senior workers who don’t yet need them.
The tasks being created by AI adoption are often highly specialized, requiring seniority and judgment. This creates a precise and dangerous bottleneck: the new jobs are inaccessible to the very people displaced from the old ones.
Think about what this means structurally. The on-ramp to a career in knowledge work has historically been: do the repetitive analytical work, learn the fundamentals through practice, build toward judgment over time. AI is removing the first step. Which means we are eliminating the mechanism through which people develop the judgment we then claim to need.
This is not a technology problem. It is a career design problem. It is an organizational design problem. And it requires an educational response - both inside institutions and inside companies - that we have not yet built.
The question again, and it keeps coming back to the same place: are we building the systems that give people a real shot at getting there?
What abundance actually requires
The optimists are not wrong. The democratization of capability that AI represents is genuinely profound. A person with no technical background can now build in an afternoon what would have required an engineering team two years ago (out of necessity, I build a healthcare / wellness-related app last week, for instance). The kid in the small town who was previously locked out of opportunity by geography and access now has tools that can genuinely change the equation.
But democratization is not automatic. It does not happen simply because the technology exists. It requires:
Foundational literacy in how to work with AI systems - not coding, but prompting, directing, evaluating. The cognitive skill to recognize when AI is wrong, which requires deep domain expertise, which requires a career track that still develops domain expertise.
Access - broadband, devices, and the cultural permission to believe these tools are for you, not just for people at well-resourced companies or specific communities.
And an education system that stops training people for an economy that no longer exists.
The World Economic Forum estimates that 39% of core workforce skills will change by 2030. Of the global workforce, roughly 11 workers in every 100 will need training but are unlikely to receive it - approximately 120 million people globally at medium-term risk of redundancy.
Not because the jobs disappeared. Because the training won’t arrive.
The only question that matters
Griffin’s Friday depression is understandable. So is the excitement of watching something historically unprecedented unfold in real time. Both are true simultaneously and neither one of them is a strategy.
The right question is not: what jobs will AI destroy?
The right question is: what does it take to be on the right side of this transition - and are we building the systems that give people a real shot at getting there?
The answer requires honesty about where the real bottleneck is. It is not the technology. The technology is moving faster than any institution can regulate or any curriculum can track. The bottleneck is human systems: education, training, organizational design, career pathways, and the political will to treat workforce readiness as the national priority it actually is.
The organizations, institutions, and individuals actively building new capability frameworks right now - not waiting for the curriculum to catch up, not waiting for policy to lead - are the ones who will write the outcome.
The bottleneck is real. But bottlenecks, by definition, can be broken.
That is the only question worth asking. And we haven’t answered it yet.
Tyler Blackwell is Director of Global Workforce Planning at Morningstar and creator of the Factors of Career Choice© framework. Layers Not Linear explores how people, organizations, and ideas actually work — through behavioral science, career design, and the economics of work.
Sources and Further Reading
Challenger, Gray & Christmas. Job Cuts Report. April 2026. cfodive.com
Challenger, Gray & Christmas. Job Cuts Report. Year-to-date data. investing.com
Oxford Economics / Fortune. “AI Layoffs Are Looking More and More Like Corporate Fiction.” January 2026. fortune.com
Forrester Research. Predictions 2026: The Future of Work. hrexecutive.com
CNBC. “AI Was Behind Over 50,000 Layoffs in 2025.” December 2025. cnbc.com
IDC Analyst Brief. “Closing the Gap: Verifying AI Skills in the Enterprise.” workera.ai
McKinsey & Company. State of AI Report. 2025. gloat.com
PwC. Global AI Jobs Barometer. 2025.
World Economic Forum. Future of Jobs Report. 2025. iternal.ai
IMF Staff Discussion Note SDN/2026/001. “Bridging Skill Gaps for the Future: New Jobs Creation in the AI Age.” January 2026. imf.org
Rezi.ai / Burning Glass Institute. “The Crisis of Entry-Level Labor in the Age of AI.” January 2026. rezi.ai
Brookings Institution. “How AI May Reshape Career Pathways to Better Jobs.” April 2026. brookings.edu