Much of the noise is coming from one place: the technology industry itself. That’s understandable. Software engineering is one of the first professions where AI has delivered real, visible productivity gains. Output that once required teams now requires far fewer people. When disruption hits close to home, anxiety travels fast.

But extrapolating from one sector to the entire economy is a mistake. We should not, and cannot, accept what’s being sold as a foregone conclusion. If we intervene now, we can meaningfully reduce the blast radius of AI disruption. The first step is separating what we know from what we don’t know, which gives us a clearer path forward.

What We Actually Know

AI is already reshaping work. That change is real, and the transformation of software engineering jobs explains much of the current panic emanating from Silicon Valley. But despite the headlines, there is still little evidence that AI productivity is the primary driver of today’s broader labor market churn. Research published in January by Oxford Economics found the evidence of an AI-driven shakeup to be patchy at best. Labor economists and AI experts at the Wharton School have argued much the same thing, citing “AI-washing” of job losses. In fact, much of what we are seeing reflects familiar forces: economic cycles, over-hiring, and cost correction. 

Unemployment rates don’t lie: the U.S. rate sits at 4.4% (9.4% for 16-to-24-year-olds) — far below EU unemployment peaks in the 1990s, when rates hit 11% overall and exceeded 20% for young workers.

We also know that most organizations trying to deploy AI are discovering that the hardest problems are not technological. Data readiness, security, integrations, workflow redesign, and building human skills remain stubborn bottlenecks for true AI implementation. According to McKinsey, two-thirds of companies using AI have not scaled it across their enterprise. That diffusion will take time — a crucial point missing from most of the current debate.

There is also growing empirical evidence that poorly implemented AI simply increases work intensity — flooding organizations with output that still requires human attention, judgment, and correction. A recent Harvard Business Review study even coined the term “AI brain fry” to describe the information and workload fatigue that comes with too many AI tools at once. Automation can increase intensity and stress rather than productivity.

The bottom line: systems aren’t ready, and people aren’t ready to use AI well. The apocalypse is likely not coming as fast as you think.

What We Don’t Know

But AI transformation is coming, in some form, across nearly all industries — and genuine unknowns remain. We don’t yet know how demand will respond as AI compresses the cost of expertise. We don’t know how quickly different sectors will adapt. And we don’t yet know which new jobs and industries will emerge — although history tells us they will. Predictions that all white-collar work will be automated within a year or two aren’t realistic. But five years? In some domains, perhaps. In most others, regulation, safety concerns, and legacy infrastructure will slow adoption dramatically and keep people in the loop. What is coming is akin to the “process reengineering” wave of the 1990s, with AI agents being embedded into workflows.

The Real Risk Is Underinvesting in People

The real risk, then, is not runaway technology displacing workers en masse. It is failure to invest in people. If we blindly accept AI as a substitute for humans, the pessimists may be proven right. But if we treat it as a force that improves and amplifies what people can do, a very different — and more prosperous — future is possible.

That requires all of us to shift our focus and realize that every positive outcome in an AI economy is reliant on our investment in human capability. In the workplace, this means moving beyond narrow task automation toward deliberate job and workflow redesign — and embedding learning directly into people’s everyday work alongside tech deployment, rather than after the fact. In education, the priority should be prioritizing learning how to learn — building AI literacy across disciplines and age levels, supporting and training teachers, and creating career pathways that extend beyond traditional white-collar roles. And at a societal level, it means recognizing that in a world where AI can generate content, credentials, and even identities, trust becomes our most valuable currency. Institutions and credentials that can reliably assess, verify, and signal human skills will matter more than ever.

The story of AI does not have to be one of inevitable displacement. The current anxiety reflects who is feeling disruption first — not where the economy must end up. The companies that grow, create new jobs, and win aren’t the ones that implement AI the fastest; — they are the ones that implement learning the fastest.

The future of work in the AI era will not be decided by what machines can do. It will be decided by what people can do — and how seriously we invest in them.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.