AI Isn’t the Threat, the Transition Is

Much of the current public conversation treats job displacement as a narrow labor issue, solvable through retraining or eventual job creation. But employment is not merely a matter of individual livelihoods; it is a core economic engine. Consumption depends on income. Public services depend on tax revenue. Social stability depends on participation and purpose. When large segments of the population are displaced faster than they can be productively reabsorbed, the effects ripple far beyond the labor market.

I am deeply skeptical that Universal Basic Income will emerge in any meaningful or adequate form. If it does appear, it is more likely to resemble welfare under a different name, insufficient in both scale and design to preserve dignity, purpose, or economic stability. Even generous transfer payments do not replace the social and psychological role of work, nor do they compensate for the loss of upward mobility and economic agency. The more likely outcome is a widening divide between those who own or control productive capital and those who do not, with predictable consequences in the form of social strain and political instability.

Recently, I listened to Jensen Huang, CEO of NVIDIA, speak at the World Economic Forum in Davos with interest, but also with caution. His optimism is articulate and, in many respects, historically grounded. He argues that AI will act as a productivity multiplier, empowering workers rather than replacing them, and that technological revolutions have always created more jobs than they destroy.

I do not believe Huang is wrong so much as incomplete. His argument makes sense over a long time horizon, and history does show that technology eventually gives rise to new industries and new forms of work. But history also shows that transitions are rarely smooth, evenly distributed, or benign in the short to medium term.

Where I differ is on speed and distribution. AI is advancing faster than labor markets can adapt, and it is displacing entry-level cognitive jobs first, the very roles that have traditionally served as on-ramps to careers. These are not abstract positions. They are junior analysts, support staff, assistants, customer service representatives, and coordinators who form the base of professional ladders. When those rungs disappear, retraining alone cannot compensate, because the pathway itself has been removed.

Even if new jobs emerge, they may not emerge quickly enough, in sufficient quantity, or in forms accessible to the same people being displaced. The assumption that labor can seamlessly transition into more advanced, AI-adjacent roles underestimates both the friction involved and the diversity of human capability. Adaptability is not evenly distributed, nor are opportunity and access.

We also underestimate second-order effects. Job displacement is not merely about unemployment rates; it affects consumption patterns, housing markets, healthcare systems, and public finances. As demand weakens, businesses contract. As tax revenues decline, public services strain. These feedback loops matter economically, even if the long-term outcome of AI adoption proves positive.

Finally, while I respect Huang’s perspective, it must be viewed through the lens of structural incentive. As the leader of a company whose success is directly tied to the rapid and widespread adoption of AI infrastructure, his framing naturally emphasizes growth, opportunity, and reassurance. This does not invalidate his position, but it does mean it should be weighed alongside assessments that are less commercially aligned and more focused on transitional risk.

The real debate, then, is not whether AI will ultimately create value. It almost certainly will. The question is whether our economic and social systems are prepared for the pace at which that value is being created, and for the uneven disruption that will occur along the way. Ignoring the transition does not make it safer; it merely delays the reckoning.