The UX of AI Itself Is Broken: 'Confident Wrongness' Destroys Trust
uxmag · x · 2026-09-15
Part 8 of UX Magazine's "UX × AI" series, by Tushar Deshmukh, turns the lens on AI products themselves: they routinely violate usability heuristics settled since 1994, and users give them exactly one strike.
The key concept is "confident wrongness" — incorrect AI answers that look indistinguishable from correct ones, so users can't judge reliability from presentation alone, and a single mistake destroys trust. The article argues the problem isn't interface-level but inherent to how AI outputs are presented: missing evidence chains, confidence signals, and uncertainty expressions undermine usability at its foundation.
More from Apps
- AI slide builders keep overriding negative prompts: Gamma users hit hardcoded layout walls — waterySoup2344 · 2026-11-02
- User boards flight booked entirely by Grok Bot — XFreeze · 2026-11-02
- Motion Prompt: a 192-effect motion dictionary that makes AI nail your animations — aigclink · 2026-11-02
- Dev builds zero-subscription calorie tracker with Gemini BYOK, no backend needed — bonfiglioalessio · 2026-11-02
- User uses Grok Bot to track stock and snags an iPhone Duo — rudrank · 2026-11-02
- AI Assistants Remember You, But Nothing Helps You Find What They Told You — Ok_Negotiation_2587 · 2026-11-02