Study: AI shopping assistants vary price info by wealth; 71% use AI for discounts
A new study reports that AI shopping assistants such as Claude and ChatGPT may show different price information depending on a user's wealth, renewing concerns about fairness and transparency in AI-driven shopping. A Snipp survey of 1,000 US adults who often use AI for search and shopping found 61.8% had bought an item after an AI recommendation and 71.3% had asked AI to find promotions, discounts or coupons, while 58.1% used AI to compare products and 41.6% treat AI as a starting point for product searches, versus 71.6% for traditional search engines. A separate test of 50 e-commerce sites and 450 real shopping tasks found Claude, GPT and Gemini completed about 80% of tasks, with return windows easiest at 91% and shipping costs or finding products and prices harder; failures came mostly from unclear information, JavaScript-loaded content or region and login limits, and half of testable product pages kept prices in JavaScript rather than the server HTML.
AI is shifting from a search tool to a shopping decision layer, making price disclosure, information readability and algorithmic fairness new pressure points for platforms and regulators.


最新研究指出,Claude、ChatGPT 等 AI 購物助理在搜尋商品時,可能會依使用者的財富狀況提供不同的價格資訊,讓 AI 在購物場景中的公平性與透明度再度引發關注。
研究同時顯示,AI 已不再只是協助搜尋資料的工具,而是逐步介入商品發現、比價、找優惠與購買決策。Snipp 針對 1,000 名經常使用 AI 工具進行搜尋與購物的美國成年人調查發現,61.8% 的受訪者曾在收到 AI 推薦後購買商品,71.3% 曾請 AI 尋找促銷、折扣、優惠券或優惠活動,另有 58.1% 會用 AI 來比較商品。