AI Shopping Guides Hit 16% Accuracy, Shanghai Consumer Council Warns Platforms Against Algorithmic Exploitation

The Shanghai Consumer Council published its '2026 618 Online Shopping Experience Survey' today, drawing on 4,308 valid responses. While 84.56% of consumers have used AI-assisted shopping features, only 16.06% found that AI accurately matched their product needs — a striking gap that underscores widespread dissatisfaction.

Nearly 39% of consumers reported that AI recommendations entirely disregard low-price preferences, instead prioritizing premium items. An additional 29.71% described results as a blend of high and low prices requiring significant manual filtering. On platform algorithm performance, just 24.21% of consumers deemed recommendations well-matched to their needs, while 58.33% received repeated pushes for previously viewed or carted items — 26.69% of whom rated the experience as poor.

More troubling are signs of algorithmic price discrimination: 38.51% of consumers observed that identical products carried different promotional terms across accounts, with over 40% of those users reporting purchase anxiety as a result. Regarding discount structures, 21.4% of respondents cited 'overly complex logic that makes final pricing incalculable', while nearly half described the rules as labyrinthine and requiring repeated verification.

AI Enthusiasm and Experience Divergence

Despite lackluster AI shopping experiences today, consumer anticipation for one-stop AI purchasing remains remarkably high: over 85% of respondents expressed positive expectations, with 38.65% eagerly awaiting one-click Agent shopping and 46.87% open to trying it in appropriate contexts. The report further indicates that the 2026 618 promotion cycle has moved beyond broad price wars, as consumption shifts from 'people-seeking-products' to 'AI-assisted selection'. Instant retail now grows at roughly ten times the pace of the broader e-commerce market.

The Shanghai Consumer Council stated unequivocally that intelligent-agent-driven shopping models hold significant promise, but warned that platforms must place consumer experience at the center — recalibrating algorithm logic to weaken commercial incentives and strengthen demand alignment in pursuit of 'algorithms for good'. As early as April, a separate Council survey found that over 90% of respondents detected information bias or algorithmic inducement in platform recommendations; two months later, the issue persists with little improvement.