AI Is Your New Doctor—But Not Your Shopper.. why 63% of UK Consumers Use Generative AI, Mostly for Health Advice?

More than six in ten UK consumers—63 percent—used a generative AI application within the past year, according to Deloitte’s Digital Consumer Trends Report (2026), marking a 17-percentage-point increase from the prior year. One in ten now interacts with AI daily.

Health Advice Outpaces Shopping Assistance

Fifty-one percent of AI users explicitly avoid using the technology for shopping and have no intention of doing so in the future. Only 15 percent rely on AI to discover products, and just 14 percent use it to generate purchase ideas. In contrast, 26 percent report turning to AI for health-related advice, 20 percent for diet or fitness guidance, and 10 percent treat it as they would a therapist.

A key divergence emerges in user expectations: a flawed shopping recommendation is immediately apparent and erodes trust outright, while medical advice—delivered with clinical terminology and personalized questioning—feels authoritative even when its accuracy cannot be verified by the user.

Algorithm Aversion Explained

Research by Dietvorst, Simmons, and Massey (2015) identified “algorithm aversion”: people abandon statistical algorithms after observing even one error—even when the same mistake occurs with a human forecaster and the algorithm still outperforms overall. This phenomenon helps explain resistance to AI in commerce: a single unsuitable clothing or appliance suggestion can permanently sever reliance on the tool.

By contrast, health queries rarely yield immediate feedback on correctness. Users lack baseline knowledge to assess diagnostic plausibility, and AI responses mimic professional tone and structure—enhancing perceived legitimacy without requiring verification.

From Dr. Google to Dr. ChatGPT

Longoni, Bonezzi, and Morewedge (2019) previously documented strong resistance to medical AI, rooted in “uniqueness neglect”—the belief that personal illness demands individualized human judgment. That resistance has receded as AI interfaces evolved to simulate tailored consultation: asking about age, symptom duration, and associated factors before delivering context-specific responses.

Deloitte’s data shows 38 percent of those seeking health advice from AI believe it is always factually accurate; among users seeking guidance on other personal matters, that figure rises to 42 percent. Yet confidence remains decoupled from reliability: AI systems are designed to project certainty, but factual errors persist at rates no developer has eliminated.

Sixty-one percent of respondents say they would be less inclined to contact customer service if they knew an AI assistant would handle their inquiry. The same survey found that users trust AI with symptoms—but not with shopping lists.