Economists’ AI Forecasts Missed the Point, Says Alex Imas
In a recent discussion that has quickly become a reference point for those concerned about the intersection of technological advancement and labor markets, Alex Imas articulated a perspective that positions the prevailing confidence of many economists regarding artificial intelligence as not only overly optimistic but potentially blind to the magnitude of disruption that AI could introduce to employment structures, thereby suggesting that the discipline’s standard analytical frameworks may be ill‑equipped to capture the full spectrum of risk associated with rapid automation.
Background
For decades, mainstream economic thought has tended to treat technological progress as a largely benign force that, through the mechanism of creative destruction, ultimately yields higher productivity, new industries, and a net increase in well‑being, an approach that has proven useful in explaining historical shifts but that now appears increasingly strained when confronted with the unprecedented speed, versatility, and generality of contemporary artificial intelligence systems, which are capable of performing tasks that were previously conceived as uniquely human and that are being integrated across sectors ranging from manufacturing to services with a velocity that outpaces many of the assumptions embedded in traditional labour‑market models.
Imas’ Argument
Imas contended that the prevailing optimism among economists rests on a set of premises that, while historically valid, fail to accommodate the scale and scope of AI‑driven automation, particularly the technology’s ability to substitute for both routine and non‑routine cognitive tasks, thereby eroding the protective buffer that previously insulated a substantial segment of the workforce from immediate displacement; he emphasized that the standard equilibrium‑based models, which often assume smooth adjustment processes and the swift emergence of new occupations, do not adequately reflect the empirical evidence of skill‑specific mismatches and the latency in creation of comparable roles, leading to a systematic underestimation of the short‑ to medium‑term employment shock that could materialize.
Moreover, Imas highlighted that the methodological focus on aggregate productivity gains overlooks the distributional consequences for workers whose skills are rendered obsolete, a flaw that is compounded by policy cycles that frequently lag behind technological adoption, resulting in a paradox where the very tools designed to enhance efficiency simultaneously generate a labor market environment where the promise of new jobs remains speculative at best and, at worst, a narrative employed to justify inaction on more pressing matters such as reskilling, social safety nets, and regulatory oversight.
Institutional Gaps and Procedural Inconsistencies
The critique extended to the institutional practices of research bodies and policy advisory groups, which, according to Imas, often rely on data sets that are insufficiently granular to capture the nuanced ways in which AI is being deployed across micro‑tasks, thereby producing forecasts that are not only overly optimistic but also detached from the lived realities of workers facing imminent displacement, a situation that is further exacerbated by the lack of a coordinated framework for monitoring AI adoption rates in real time, a procedural omission that permits a disconnect between academic projections and on‑the‑ground developments.
Imas also pointed out that the conventional reliance on historical analogues—such as the mechanization of agriculture or the introduction of assembly‑line production—fails to account for the fundamentally different character of AI, which does not merely replace physical labor but also encroaches upon decision‑making, analytical, and creative processes, a divergence that renders the analogy insufficient and, in effect, a methodological blind spot that perpetuates policy inertia.
Implications for Policy and Future Research
If the concerns raised by Imas are heeded, the implication is that a substantial recalibration of both economic modeling and public policy is required, encompassing the integration of dynamic, non‑equilibrium approaches that can accommodate rapid technological diffusion, the establishment of real‑time labor‑market intelligence systems capable of tracking AI‑induced task displacement, and the proactive design of educational and training programs that anticipate skill demands rather than reacting to them after the fact, a suite of measures that, while demanding considerable coordination, would address the systemic shortcomings identified in the current forecasting paradigm.
In sum, the argument presented by Alex Imas serves as a stark reminder that the confidence with which many economists have approached the AI revolution may be misplaced, not because the technology lacks transformative potential, but because the analytical tools, institutional practices, and policy frameworks that have traditionally guided economic forecasting are ill‑suited to capture the multi‑dimensional risk profile that AI introduces to the world of work, thereby underscoring the urgent need for a more critical, data‑driven, and forward‑looking approach to understanding and mitigating the labor market challenges that lie ahead.
Published: April 19, 2026