ADERERROR's FW26 collection, called "Time After Time," is organized around pottery. Not as decoration. As method. Clay-inspired tones, rounded silhouettes, textures that reference the handmade. The collection argues that craft process is the content, that the way something is made is visible in the finished thing. This week, researchers at arXiv published OpenDiscoveryTrace, a new benchmark for evaluating AI scientists that criticizes existing tests for only measuring final outputs, the code, the hypothesis, the result, rather than the process that produced them. The fashion collection and the academic paper are making the same argument about what quality actually is.
Why Process Is the Thing Being Lost
The pottery metaphor is useful because pottery makes process visible in a way that most manufacturing has eliminated. You can see where the hands were. ADERERROR is selling that legibility back to a market that has spent a decade buying things made by processes it cannot see or trace. The OpenDiscoveryTrace paper is responding to the same erasure in AI research: a model that produces a correct answer by hallucinating its way through the reasoning looks identical to one that actually understood the problem. Current benchmarks cannot tell the difference. A 2025 paper in Nature Machine Intelligence by Bommasani et al. made a related point: that evaluating AI on outputs alone systematically rewards fluency over understanding, which is approximately the critique a craft educator would make of a student who copied a good pot without learning to center the clay.
The Suzhou Museum and the Architecture of Process
The Suzhou Museum of Contemporary Art, designed by BIG and organized as twelve pavilions reimagining traditional Chinese garden architecture, opened this week. It is 60,000 square meters built around the idea that the path through a space, the process of moving through it, is inseparable from what you experience at each destination. ADERERROR's pottery, the AI benchmark problem, and the Suzhou building are all making the same structural claim: the journey is part of the output, and any evaluation system that strips it away is measuring the wrong thing. Eugene Whang, who spent twenty years with Jony Ive, has described design as a practice where the reasoning behind a decision is as important as the decision itself. That reasoning is exactly what current AI evaluation, and most fast fashion supply chains, have learned to discard.