Deadly AI Sins for UX Professionals – NN/G (www.nngroup.com)

NN/g outlines a practical guide for UX professionals: seven “deadly AI sins” observed across ~400 hours of conversations with 3,000 practitioners—and the seven virtues to counter them. The sins are outsourced thinking, wasted time, lost details, isolated ideation, naïve trust, bland taste, and a defensive outlook; their counterparts are ownership, automation, selectivity, inclusion, skepticism, originality, and experimentation. The piece urges UXers to think first and use AI as a partner, automate repetitive work with reusable prompts or custom GPTs, verify outputs (to avoid hallucinations), keep stakeholders involved in ideation, and guard against generic AI-produced designs by adding human originality. Significance for the AI/ML community: this is an operational manifesto for integrating LLMs into product work without eroding expertise or accountability. Technically, it emphasizes cost/benefit patterns—invest in AI where tasks repeat (bots/custom models) rather than one-offs, maintain provenance and verification pipelines for model outputs, and design workflows that preserve the human decision-making loop. NN/g’s concrete example—training a custom GPT to draft exam questions—illustrates measurable ROI while still requiring human curation. The article’s guidance has immediate implications for tool builders, orgs, and researchers aiming to embed models into UX workflows responsibly and effectively.