Discovery isn't a phase or operational loop — it's a judgment capability built through double-loop learning: documenting reasoning before decisions and reflecting after outcomes to convert experience into compoundable judgment. AI accelerates execution but cannot develop human judgment, which remains the only advantage that grows through use rather than update
NNG: Your New UX Habit - Establishing Baselines for Impact
Gather baseline metrics before starting a project so your team can demonstrate its impact
AI: The Magic 8-Ball vs. Gen AI - a surprisingly interesting comparison
A surprising comparison between the Magic 8-Ball and generative AI: both sample from distributions, but opposite design contracts — one says "I'm a guess" with honest uncertainty (plastic, $2), the other says "I'm an answer" with fluent prose hiding probability (massive infrastructure). The design challenge for modern AI is to borrow the 8-Ball's honesty (surface uncertainty, cite sources, allow refusal) while keeping fluency and convenience
Prototyping: I translated user behavior into 184 UI decisions
A "Behavioral Translation Dictionary" translates user conditions (e.g., high anxiety) into design decisions through a chain: Context → Need → Rule → Interface Decision (35 patterns, 184 decisions total). It makes design reasoning defensible and traceable — shifting from "I think it looks better" to evidence-based logic
Opinion: Discovery research is not dead. It might be becoming even more relevant
When AI makes building cheap, discovery becomes more critical, not less — it acts as a filter, not a bottleneck, deciding what's worth testing before you build. AI mines what you already know but is blind to unknown needs, and testing every idea with real users costs time, fatigue, and product bloat
@uxdigest
