Conlon and Schwardmann ran three incentivized market experiments with 1,025 US buyers and sellers on Prolific. Pure-AI games out-sold both amateur and professional human-designed games, and labelling games "human-designed" did not raise what buyers were willing to pay.
Why this hits
The "crafted by humans" line sits in half the positioning decks I read. It is a story the team tells itself. In this market, the buyers did not pay for it.
In the Blind arm the AI premium was 6.8 cents. In the Disclosed arm, where consumers knew the game came from an AI, the premium was 6.0 cents. The difference between the two arms came out at p=0.79.
Showing the "human-designed" sticker to buyers moved the price by less than a cent.
The producer side is sharper. Amateurs expected to beat the AI by 9 cents. Professionals expected to beat it by 17 cents. The forecast only flipped after they read what the AI had shipped.
The scene I keep seeing
A head of brand at a B2B software company, mid-market, in a pricing review last month. She walks the room through why their enterprise tier carries a thirty percent premium. Her second slide reads "built by senior designers, not generated." It is her anchor.
Nobody at the table asks the obvious question. Does your buyer pay for that line, or does your sales team?
In most reviews I run, the honest answer is that the premium is defended by the people selling it and bought by very few people on the other side. The "human" story survives because the room shares the same forecast error the paper measured: a half-standard-deviation belief in a human edge, held before anyone has compared the output.
Keep the caveats visible. This is a game-design market on Prolific, not haute cuisine or senior-partner advisory. The "professionals" here are mid-career Prolific creatives. Luxury, provenance and community-driven categories may still carry a human premium. The paper tests pricing in a market, not identity in a brand.
For a lot of ordinary creative output that executives buy and sell every day, this paper settles one of the load-bearing assumptions behind the "AI will not replace creatives" line. The authors are blunt about it.
"We find no evidence that consumers intrinsically prefer human-made products."
Before the next creative budget cycle, run a blind preference test on your own category with a hundred buyers. Look at the result before your team tells you how they would price against the AI.