# Design Systems in the AI Era: Competency vs. Craft ## Lingering Questions - Where exactly is the line between "measurable" and "judged" — is accessibility's writing-level/style criterion actually competency or partly differentiator, and does that ambiguity matter operationally? - Can AI evaluation tooling, given enough training data on "craft" judgments (design critiques, dark-pattern audits), eventually approximate taste — or is there a principled reason (not just current capability) it can't? - What does "stewardship" look like operationally at organizations too small to have a dedicated design-system team — does the argument scale down? - How do you measure "convergence" (everyone's AI output looking the same) at an industry level, and is there early empirical evidence for it yet, or is this still anecdotal? ## Article Ideas - A competency-vs-differentiator self-audit checklist for a WordPress design system (patterns library, theme.json tokens) — where is the work AI-assistable vs. judgment-requiring? - A short piece translating Hunt's "repository + people" framing into a WordPress Core Design/Core AI governance question: who stewards theme.json and pattern-directory standards over time, not just who maintains the repo? ## Cross Reference - [[3. Resources/Permanent Notes/Outsourcing Judgment Removes the Accountable Decision-Maker]] — same underlying worry (delegated judgment erodes accountability) from a different domain (marketing/knowledge-work vs. design systems). - [[3. Resources/Permanent Notes/The Bottleneck Moved From Code Generation to Review and Judgment]] — parallel "generation got cheap, judgment is the constraint" pattern in engineering. - [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] ## Fiction & Worldbuilding None. ## Vocabulary - **Kano model**: satisfaction-mapping framework (must-haves, delighters, indifferents) - **Competency markers**: measurable UX floor — consistency, accessibility, performance, reliability - **Differentiators**: judged UX ceiling — personality, craft, empathy, respect - **Stewardship**: ongoing human ownership of a system's evolving standards ## Bib capsule Michelle Hunt (director of UX engineering, Lone Rock Point), "Design Systems in the AI Era: Competency vs. Craft," WordCamp US 2026 (YouTube talk, AI track). A conceptual, non-technical talk arguing AI raises design-system "competency" (the measurable floor) but cannot raise "differentiation" (the judged ceiling) without active human stewardship. ## Author's thesis **Q1 — What is this about?** How AI is reshaping design-system work, and why AI's strength at raising measurable quality floors (consistency, accessibility, performance, reliability) doesn't translate into raising the ceiling that makes a product distinctive. **Q2 — How does she develop it?** Hunt anchors the talk in the Nielsen Norman definition of a design system ("a set of standards... reducing redundancy while creating a shared language"), splits design-system work into four AI-assistable "competency markers" versus four human-judged "differentiators" (personality, craft, empathy, respect), argues that even best-case AI output converges toward generic "fine but not good," and closes by returning to NN Group's neglected second half of that definition — the people managing the repository — proposing a continuous Lean UX-style assume/build/measure/learn/adjust loop as the mechanism for ongoing human stewardship. ## Key arguments - Competency work (AI-assistable: tokens, accessibility audits, performance auto-optimization, AI code review) only lifts the floor; the ceiling stays capped by the same short-sighted decisions as before, and expectations keep rising, so standing still is a losing position. - Differentiation — personality, craft, empathy, respect — is judged, not measured, and requires human taste; competency metrics are pass/fail, differentiation isn't. - A design system without active human stewardship (the "people managing it" half of NN Group's repository+people definition) isn't really a design system — it's an artifact heading toward obsolescence, because "a machine can produce things and a machine can evaluate things, but it can't be responsible." ### AI lifts design floors, not ceilings > [!quote] ID:rw_atom1 > "we're kind of only lifting the floor... The ceiling is still being impacted by the same short-sighted decision-making as before." Mid-talk, right after the four competency markers section — the core mechanism explaining why AI-assisted design-system work alone doesn't produce differentiated output. ### Differentiation is judged, not measured > [!quote] ID:rw_atom2 > "differentiation is judged not measured... judgment takes taste and experience and opinions." The talk's central axis-defining claim, delivered right after the personality/craft/empathy/respect breakdown; everything downstream (stewardship, the feedback loop) follows from this split. ### Machines can't be responsible for design systems > [!quote] ID:rw_atom3 > "a machine can produce things and a machine can evaluate things, but it can't be responsible." The stewardship section's load-bearing claim, tied to the Nielsen Norman "repository + people" framing — also the talk's most contestable claim (flagged in the source's own Logical Validity pass as asserted rather than argued for). ## Permanent candidates - Atom 1 + Atom 2 together warrant a Permanent Note: the competency-vs-differentiator axis is a genuinely reusable framework for evaluating any AI-assisted creative-work pipeline, not just design systems. - Atom 3 warrants its own Permanent Note: the stewardship claim ("repository without people isn't a system, it's an artifact") is a distinct, portable governance claim. ## Processing checklist - [x] Read source and pulled verbatim atoms - [x] Checked existing Permanent Notes for duplicates - [ ] Register in relevant MOC(s) — orchestrator handles centrally - [ ] Run backlog.py — orchestrator handles centrally