Skip to content
Article Nielsen Norman Group Oct 2026

NN/g: AI makes UX work faster, but practitioners have mixed feelings

Nielsen Norman Group surveyed 521 UX practitioners between August and September 2026 to measure how AI is changing their work. The results, published on October 23, offer a detailed picture of where AI has genuinely landed in design practice — and where it has not.

Access is nearly universal: over 99% of respondents had AI tools at work. Two-thirds (66%) had used at least one agentic tool, with startups and small teams at 85% and nonprofits and government agencies at 36–42%. Operations and product management roles reported the highest adoption (81–83%); content design and research were lowest at 35% and 57%.

Where AI actually helps: code writing, desk research, and analyzing research data were the most common uses and also the tasks where practitioners saved the most time. User-facing work remained the exception — 61% of respondents said they had never used AI for interviewing users, and testing designs with users was rarely supported by AI.

Speed savings are real but absorbed. The average time-savings rating was 4.2 out of 5, yet the overall sentiment about AI at work was only 3.57 out of 5 — mildly positive. Among the 226 respondents who left open comments, 69% expressed at least some negative sentiment, compared with 20% who were purely positive.

The frustrations are consistent: unreliable output, the effort of checking and fixing it, mandated use from management, and the experience of cleaning up AI-generated work that others produced and left incomplete. The main praise is speed, with some practitioners — especially solo designers and small-team members — reporting new capabilities alongside efficiency gains.

On job outlook, 65% expected their role to stay important or grow, but 18% expected it to shrink or disappear. Researchers were more pessimistic than designers (23% vs. 12% expecting role reduction). Job security ratings were lowest for designers (2.99/5) and researchers (2.94/5), highest for product managers (3.81/5) and government workers.

The survey’s central finding is methodological rather than motivational: time savings at the task level do not reliably translate to system-level productivity gains, because organizations raise output expectations to match and practitioners spend significant time reviewing AI work. The authors recommend measuring productivity at the team or project level rather than by individual task speed.

For UX teams trying to set realistic expectations about AI adoption: the survey is a useful counterweight to vendor claims. AI is in the workflow, it does save time on specific tasks, and most practitioners believe their role will endure — but satisfaction with AI tools remains modest and the quality-checking overhead is real.