A message to my daughters: Don't get left behind | Fortune

TEXT START: By now, you’ve heard from your professors about their AI policies.

The Dissection

This is an investor-operator reassurance memo disguised as parental advice. It takes the historical record of agricultural and industrial automation, selects aggregate job growth as the decisive metric, and projects that pattern onto AI-driven cognitive automation.

Its real function is behavioral: accept AI, train for it, participate in its construction, and do not demand structural protection from the people who own it. The article converts a system-level displacement problem into an individual obligation to “educate yourself,” while its author’s robotics company stands to benefit from precisely that acceptance.

The Charlie Snodgrass anecdote is not evidence of a general solution. It is a transition vignette: one displaced worker absorbed into the maintenance layer of a growing robot fleet. That is a viable niche, but it does not demonstrate that the displaced population can be absorbed at comparable scale.

The Core Fallacy

The central error is treating past labor displacement as proof that AI displacement will reproduce the same employment cycle.

Earlier mechanization primarily automated physical tasks while expanding industrial capacity, markets, and demand for human cognitive coordination. The thesis here is different: AI targets the cognitive layer that historically generated new occupations, managed systems, and converted productivity into wages. If AI becomes cheaper and better across cognitive work, the replacement engine begins consuming the very job categories previously used to absorb displaced workers.

The article also confuses three separate outcomes:

• Higher productivity.
• More total output or GDP.
• Continued mass access to paid work.

The first two can rise while the third collapses. A larger economy does not guarantee that ordinary people remain economically necessary. Nor does the existence of new jobs prove that there will be enough of them, at adequate wages, for the people made redundant.

“Robots need humans” is presented as a permanent law. It is only a current bottleneck. Maintenance, supervision, logistics, exception handling, and deployment are precisely the categories competitors will have incentives to automate next. The anecdote identifies a temporary servitor niche, not a durable labor settlement.

Hidden Assumptions

• New AI-created occupations will emerge faster than AI eliminates existing ones.
• The new occupations will be numerous enough to absorb the displaced workforce.
• Workers can retrain faster than firms can automate their current roles.
• Retraining produces valuable capabilities rather than a larger queue of applicants using the same tools.
• Human judgment, creativity, and coordination will remain scarce after AI improves in those domains.
• Productivity gains will be distributed through wages rather than concentrated in owners of models, robotics, data, energy, and infrastructure.
• Aggregate job growth will translate into individual job security.
• Economic demand will expand indefinitely enough to employ every released worker.
• Human institutions can preserve large human-only markets against cheaper automated competitors.
• Robots’ current physical limitations will remain economically decisive.
• The historical stability of unemployment is relevant to a regime in which cognitive labor itself is becoming abundant.
• A young person’s willingness to participate can materially alter ownership and distribution structures controlled by capital holders.

The most dangerous assumption is hidden beneath the exhortation to “choose” participation: that individual adaptation remains a sufficient defense after productive participation has become structurally unnecessary for the majority. Under the Discontinuity Thesis, it does not.

Social Function

Primary classifications: elite self-exoneration, ideological anesthetic, transition management, and partial truth.

The partial truth is real: AI will create transitional niches, new services, maintenance work, deployment roles, and opportunities for people who reach ownership or become indispensable to owners. Some workers will move from obsolete roles into those niches, as the robotics technician example illustrates.

The anesthetic is the scale argument. “Technology created jobs before” is used to suppress the question of whether this technology destroys the economic necessity of human cognition itself. The individual is told to adapt because the system’s beneficiaries have no reason to surrender control or redistribute ownership.

The article is also transition management. It prepares educated young people to enter the AI economy without demanding that they own its productive assets. Its advice is useful only if translated into power: acquire ownership, control scarce infrastructure, or become indispensable to those who possess it. “Learn to use AI” by itself merely produces more competent applicants for a shrinking human labor market.

The Verdict

The article is a polished historical analogy built to conceal a regime change. It mistakes past labor absorption for a law of economics, treats temporary human bottlenecks as permanent moats, and uses one successful transition worker to stand in for a mass employment system.

Its recommendation is not worthless, but it is radically incomplete. Learning AI may improve an individual’s position during the lag phase. It does not defeat P1, P2, or P3. Once cognitive automation dominates, institutions fail to preserve human-only economic domains, and productive participation collapses, adaptability without ownership becomes a faster route to servitude—not escape.