Relational Goods AI Substitution Premium: A Narrow Human Hedge
Humans aren't horses, but the premium for being human only holds while humans stay scarce. The hedge is narrow: controlled-scarcity luxury and origin-priced services, not mass 'human touch.'
Chicago economist Alex Imas argues in his April 2026 essay What will be scarce? that people are not horses, because part of what we buy from each other is the fact that a human was in the loop. He calls it the “relational sector,” and Fortune flagged it as a rare optimistic thesis in the AI displacement debate.
What the research actually shows
The empirical anchor is art. In a labeled-art experiment summarized by Columbia Business School, identical work tagged “AI-generated” was valued about 62% below the same work tagged “human-made,” and Bellaiche et al. replicated the preference across multiple stimuli. The premium is real when the human origin is visible.
It is also fragile. Stanford GSB reported that when AI work flooded a stock-image platform, total sales rose roughly 39% while human artists’ revenue fell. Authenticity holds price. Supply still moves volume.
Where the premium holds, and where it does not
| Category | Holds the premium? | Why |
|---|---|---|
| Controlled-scarcity luxury (Hermès HESAY, Ferrari RACE, LVMH LVMUY) | Yes | The origin is the product |
| High-touch wealth, therapy, coaching | Yes | The human presence is the service |
| Mass “human touch” at scale (call centres, tutoring farms) | No | Origin is invisible at the margin |
| Stock illustration, generic copy, basic translation | No | Output was bought, not origin |
In the Dwarkesh transcript, Imas defines the relational sector as goods “where the fact that a human was in the loop is part of the value of that product.” UNESCO reads this through Baumol: tasks machines cannot substitute by definition keep getting more expensive relative to everything else.
What this does not tell you
The “premium disappears under mass production” claim is a logical extrapolation, not a measured result. The Stanford finding and the survey work catalogued by SCIRP show price pressure and volume substitution for unlabeled or mixed work, not a clean collapse of clearly-labeled human work. Imas’s framework is a thesis, laid out in his own essay and the companion podcast, not a settled empirical finding.
FAQ
So which businesses benefit?
Categories where authenticity is the product (controlled-scarcity luxury, in-person services priced on origin) keep their premium. Categories sold on output (stock content, generic copy, scalable tutoring) do not, and the more AI improves, the more that gap widens.
Disclaimer. This article is research commentary, not investment advice. The named tickers illustrate a thesis about authenticity-priced businesses, not a recommendation to buy or sell any security.
Past positioning of our funds does not predict future allocation, and the views above are subject to change as new evidence arrives.
