Jakob Nielsen Lists 76 Open Research Questions in AI Usability

Seventy-six. That's the number of open research questions Jakob Nielsen, the usability researcher behind the Nielsen Norman Group, says still stand between the AI industry and a genuinely usable product category.

Nielsen, who has spent roughly four decades turning interface research into practitioner guidance, published the list under the title "76 Open Research Questions in AI Usability" on his Jakob Nielsen on UX platform. The piece is not a survey of answers. It's a map of gaps — the questions nobody has yet resolved with evidence rather than opinion.

What does a question list actually signal?

When a discipline's founding figures start counting what they don't know, that count is itself data. A mature field — web usability, say — has heuristics, validated patterns, and thousands of tested interfaces behind it. AI interaction design doesn't. Nielsen's 76 questions serve as a rough audit of that shortfall.

The framing matters for practitioners. If you build AI features today, you're working largely without the equivalent of Nielsen's own 10 usability heuristics, which since 1994 have given web teams a shared checklist. AI usability has no comparable canon yet. Every one of those 76 questions is a place where a design decision currently rests on intuition instead of a measured user outcome.

Why count them at all?

Because prioritization needs a denominator. A single vague worry — "we don't understand how users trust AI" — doesn't help anyone allocate research budget. Seventy-six discrete questions can be sorted, assigned, and, eventually, checked off.

It also sets a baseline. When someone re-runs the count in a year or two, the delta becomes a progress metric for the whole field. Did the list shrink because studies answered questions? Or did it grow because researchers found new unknowns? Either result is informative, and Nielsen's numbering makes the comparison possible.

What's at stake for design teams?

The practical risk is familiar to anyone who watched early web design: teams ship interfaces tuned to what the technology can do, not to what users can figure out. Nielsen's career-long argument — that usability problems are measurable, fixable, and cheap to fix early — applies with full force to chat boxes, prompt fields, and AI-generated output screens.

Seventy-six open questions mean 76 places where a team's untested assumption could be costing task completion, trust, or return usage. The list is effectively a research agenda handed to the industry for free.

Whether academic labs, corporate research teams, or Nielsen's own network pick up the agenda first, the number to watch is the count itself — and whether next year's version of this list runs longer or shorter.