Amazon's new Alexa-powered merch generator lets you describe a vibe, print it on a hoodie, and call it creativity. Simultaneously, a 2026 paper in arXiv CS.CY by Wenjun Cao titled Generative Models Erode Human Temporal Learning Through Market Selection argues that this exact workflow is quietly dismantling the knowledge infrastructure that made good design possible in the first place. The two stories are not separate events. They are cause and consequence running in parallel.
The Market Selection Problem in AI-Powered Fashion
Cao's paper frames the danger not as AI producing bad output, but as AI training markets to select against slow, accumulated human skill. When Amazon surfaces an AI-generated graphic that gets purchased over one a designer spent three weeks on, the market signal tells the next designer: don't spend three weeks. This is precisely the dynamic that brands like Seya, a one-woman label built on years of world travel and handcraft research, are structurally unable to win inside. Keiko Seya's anti-corporate garments represent the temporal depth Cao warns we're losing: knowledge that can only accumulate through lived, embodied iteration. You can't Alexa that.
What Custom Merch Culture Actually Kills
The irony is that Amazon's feature arrives just as the art world is reckoning with its own version of this dynamic. Artnet's investigation into AI and gender inequity in the art world found that generative tools are being adopted unevenly, concentrating access for already-advantaged practitioners while mid-career women, who already face structural exit pressure, are least likely to be repositioned as AI beneficiaries. Fast Company's Natalie Nixon put it bluntly: creativity is currency, but only when it's legible to the market. AI-generated merch is legible. Decades of craft knowledge is not. The founders building AI creative tools should note that TurboFund's seed-stage AI investor list tracks exactly which VCs are backing this wave, and the pitch decks are not leading with what gets lost.