The Laws of UX, applied to mobile and AI

Laws of UX collects the heuristics and psychological principles that quietly shape every good interface. They predate generative AI by decades, but AI experiences — especially on small screens — live or die by them. The model can be brilliant and the product still fail one of these laws. Here are the ones that matter most, with how they play out in AI and mobile.

Jakob's Law

Users spend most of their time on other products, so they expect yours to work the same way. An AI feature that invents novel interaction patterns fights muscle memory and loses. Example: a member-facing assistant that hides its input in a non-standard place, or buries "undo" three taps deep, frustrates people who expect chat to behave like every other chat they use. Make the AI feel familiar; spend your novelty budget on results, not controls.

Hick's Law

More choices means longer decisions. The best AI experiences reduce choices rather than add them. Example: instead of a help screen with twenty options, the assistant surfaces the one most-likely next action — "Draft the renewal email?" — with the others a tap away. Used well, AI is a Hick's Law machine: it collapses a menu into a suggestion.

Doherty Threshold

Keep response under ~400ms and engagement holds; exceed it and attention drifts. AI is slow by nature — a complex answer takes seconds. Example: a blank spinner for eight seconds feels broken; streaming the answer token-by-token, with an instant acknowledgment, feels alive even though the total time is the same. Perceived latency is a design decision, not just an engineering one.

Tesler's Law (Conservation of Complexity)

Every system has irreducible complexity; the only question is who absorbs it. Good AI products absorb complexity so the user doesn't. Example: "prep the board packet" hides a dozen steps — pulling financials, reconciling figures, formatting, checking. The member coordinator sees one button and one reviewable draft. The complexity didn't vanish; the system ate it. This is the entire promise of designing intelligence for humans.

Miller's Law

People hold roughly 7±2 chunks in working memory. AI output that arrives as a wall of text breaks this instantly, and it's worse on a phone. Example: a long analysis becomes a three-bullet summary with "show detail" affordances. Chunk it, lead with the answer, let people expand on demand.

Fitts's Law

Time to acquire a target depends on its size and distance. On mobile that means thumb-reachable, generously sized touch targets. Example: because people will constantly accept, edit, or undo what the AI proposes, those controls need to be big, close, and unmistakable — not a 12px link in a corner.

Aesthetic-Usability & Peak-End

People perceive attractive interfaces as more usable, and they judge an experience by its peak and its end. Example: an AI flow that ends with a clean, correct, confidently-presented result leaves a far better memory than one that ends with a hedge or an error — even if the middle was identical.

AI doesn't replace UX laws — it raises the stakes on them. A model can be state-of-the-art and the experience still fail Jakob, Hick, or Doherty.

This is the heart of "designing intelligence for humans": the intelligence is necessary, but the laws of how humans actually perceive, decide, and remember are what make it usable. We build to both.