Companies are measuring AI adoption by tokens consumed. Luis Berumen Castro calls this 'tokenmaxxing' in his piece for UX Collective, a vanity metric disconnected from OKRs or user outcomes. The lead article argues designers need adaptive, water-like flexibility rather than rigid AI mandates, while Adi Leviim's companion piece makes a sharper claim: AI products have erased 20 years of empty-state design research in a single product cycle.
The 'Make Me Think' selections carry the harder technical argument. O'Reilly's piece states it plainly: organizations without automated tests, CI/CD pipelines, and documentation failed at microservices and will fail with AI coding agents for the same reason. The infrastructure deficit predates the tool. Robin Moffatt's piece on AI slop quantifies the social cost differently, framing machine-generated content as a slow kill of online communities rather than a sudden collapse.
The full edition is worth reading for three specific tensions it surfaces without resolving: whether AI returns discovery work to designers or just reframes productivity as ambition, how typeface safety intersects with AI-generated interfaces, and what Hiroshi Sato means by a one-dimensional pipe connecting two high-dimensional minds. That last one alone earns the click.
[READ ORIGINAL →]