
The introduction of AI in retail has led to a significant gap in consumer trust.
Consumers determine their trust in AI services based on task risk.
Retailers need to establish seamless transition mechanisms between AI and human customer service.
As the peak shopping season of 2026 approaches, AI agents are increasingly being deployed in the customer service frontlines of the retail industry. However, according to two recent industry studies released by cloud communications platform Sinch, there is a significant gap between the pace at which retailers are adopting AI and the level of trust consumers have in it.
Consumers determine their trust in AI based on "task risk."
Retailers are actively introducing AI, but consumer confidence is declining.
Sinch’s report on the "AI Production Paradox" points out that 63% of retail brands have officially launched AI customer communication agents. At the same time, a survey of 2,501 consumers across eight countries conducted for the "2026 Black Friday and Cyber Monday Consumer Survey" shows that only 43% of global respondents believe AI can make their holiday shopping easier, a 5 percentage point decrease compared to the same period last year.
In the United States, which has the greatest skepticism about AI, this percentage falls to 26%. Additionally, the proportion of consumers who express greater trust in human advice over AI has slightly increased to 53%, while those with no concerns about AI have decreased from 9% last year to 7%.
Despite the overall decline in optimism, the survey found that consumers are not outright rejecting AI—only one out of 2,501 respondents stated that they would not interact with AI under any circumstances. Consumers' acceptance of AI depends on the level of risk involved in the specific task:
• High trust tasks (information inquiries): In the U.S. market, as many as 76% of consumers believe AI can effectively handle "order tracking," and 64% trust AI to help understand "product information."
• Low trust tasks (financial transactions and personal data changes): When sensitive operations are involved, trust drops significantly, with only 58% believing that AI can handle "account changes," and the trust that AI can handle "payment and billing issues" is even lower at just 55%.
The report analysis indicates that while consumers are currently willing to accept AI for "reporting order status," they maintain high caution regarding allowing AI to "change settings involving money and personal data."
The "post-sales phase" becomes the key to success or failure in holiday retail.
From placing Black Friday orders to the December Christmas gift deliveries and the return and exchange wave in January, "after-sales service" is the stage where retailers spend the most time interacting with customers during the holidays, and it is a critical battlefield for determining brand reputation.
Data shows that 93% of consumers believe that logistics delivery updates are crucial to the shopping experience, and similarly, 93% of respondents stated that being able to directly "reply" to order confirmation or logistics notification messages is very important. Currently, 49% of retailers are deploying AI agents in customer service.
However, the challenge faced by retailers is the dynamic transition of conversation contexts: a customer may first inquire about the package location (an area where AI excels), followed by reporting "duplicate charges" (an area with the lowest trust in AI). If the AI agent fails to promptly identify and seamlessly transfer the situation to human customer service, any mistakes will lead consumers to blame the brand itself.
Privacy and data security become internal and external concerns.
In terms of risk management, privacy and data usage issues are the top concern for 43% of consumers worldwide. Internally, among retailers, "the risk of customer data breaches" is also the primary reason for withdrawing (Rollback) AI agent deployment (28%), followed by the AI generating "hallucinations" or responses that deviate from brand tone (18%).
In anticipation of the upcoming December logistics peak and future long-term planning, Julia Fraser, Sinch’s Executive Vice President for the Americas, proposed six practical recommendations:
Start with high-trust scenarios: Prioritize applying AI to order inquiries and product Q&A, and gradually expand functions once customer goodwill is established.
Early transfer to humans for financial transactions: Set clear triggers for sensitive conversations regarding payments, billing disputes, or personal data, and immediately involve a human agent.
Maintain AI identity transparency: Proactively inform consumers that they are conversing with an AI agent and clearly explain its scope and limitations, which helps to reduce market concerns.
Implement two-way interaction notifications: Ensure each logistics or order push notification has a reply function, avoiding customer influx to the service hotline due to the inability to directly ask questions.
Design seamless transfer mechanisms in advance: 52% of consumers are willing to use voice AI agents if there is a seamless transfer to a human agent; during transfers, it is essential to fully hand over order details, historical conversations, and the steps AI has already attempted.
Embed data safeguards into the underlying architecture: Deploy data masking and other security protections directly at the communications infrastructure level instead of repeatedly developing them each time a new AI agent is launched, preventing data breach risks from the source.
As consumers have clearly outlined the blueprint for AI trust in retail through their behaviors, the ability to achieve a precise balance between automated efficiency and human warmth will determine whether retail brands can win the long-term trust of customers in this holiday shopping season.
This article is reproduced from Hong Kong Economic Times, with all copyright reserved by the original author and media.
免责声明:本文章仅代表作者个人观点,不代表本平台的立场和观点。本文章仅供信息分享,不构成对任何人的任何投资建议。用户与作者之间的任何争议,与本平台无关。如网页中刊载的文章或图片涉及侵权,请提供相关的权利证明和身份证明发送邮件到[email protected],本平台相关工作人员将会进行核查。