Ping Identity’s Cayla Curtis on Retailers Building Consumer Trust in AI Shopping Agents: The RetailTouchPoints.com Q&A
Key Takeaways
- Consumers are willing to use AI to help them shop, but most still want to approve important decisions and stay in control before trusting AI to buy things for them.
- Retailers can build trust in AI shopping by being clear about how it works, giving customers control over permissions, and protecting their identity information.
AI shopping assistants are getting smarter, but consumers are not ready to let them make every decision. New research from Ping Identity found that 56% of consumers feel most comfortable using AI for help and advice, such as comparing products or finding a smaller number of choices. Only 31% would let AI make simple decisions on its own, such as choosing products.
People are most concerned when money and personal information are involved. On average, consumers would allow an AI agent to make routine purchases worth up to $200 without asking them first. However, 42% would set that limit at $100 or less, and 22% would not trust AI to handle any financial transaction. Trust in the companies that manage personal identity information is also falling. Only 12% of consumers fully trust organizations to manage their identity data, down from 17% last year. Also, 71% worry that an AI agent acting for them could give others the wrong idea about their personality, ethics, or values. For retailers, this means they must earn the right to let AI make decisions instead of assuming customers will accept it.
These concerns are important for retailers investing in agentic commerce. This is a type of shopping where AI agents do more than suggest products: they can also complete purchases. Retail TouchPoints spoke with Cayla Curtis, Senior Director of Product Management for CIAM and Consumer Agents at Ping Identity. She explained where consumers draw the line on AI shopping agents making decisions, how retailers can manage access to personal data and explain AI use without making shopping harder, and what they can do now to prepare for future rules about AI, identity, and data.
RetailTouchPoints.com: Your survey found that 56% of consumers feel most comfortable using AI for help and recommendations, while 31% would let AI make simple decisions, such as choosing products, on its own. What does this difference mean for retailers investing in AI shopping assistants and agentic commerce?
Cayla Curtis: There is a real opportunity here, but retailers need to understand that consumers see a big difference between AI helping them make a decision and AI making the decision for them. Ping's research found that people become less comfortable as AI moves from offering help and suggestions to taking action on its own.
For now, retailers should use AI to make shopping easier. It can compare products, narrow down choices, suggest useful items, and prepare purchases while leaving the final decision to the customer. As shoppers become more comfortable letting AI handle everyday tasks, retailers can give their AI tools more responsibility. But customers must be allowed to build that trust over time. They need to know what an AI agent can do, what information it can use, and when it will ask them for permission.
RetailTouchPoints.com: Consumers set an average limit of $200 for routine purchases an AI agent can make without asking first. However, 42% set that limit at $100 or less, and 22% would not trust AI with any financial transaction. How should retailers design automatic checkout and approval steps to respect these limits?
Curtis: There should not be one spending limit for everyone. The fact that consumers choose very different limits shows that they want to set their own rules.
Retailers should give customers detailed controls so they can decide how much an AI agent can spend, which kinds of products it can buy, and when it must ask for permission. For example, a customer might be happy to let an agent automatically buy a $30 household item they use regularly but want to approve a new $300 electronics purchase themselves. Customers should also be able to review, change, or cancel these permissions easily. If a purchase does not fit the customer's rules, the agent should smoothly hand control back to the customer and ask for approval.
How Can Retailers Build AI Shopping Agents That Reflect Customer Values?
RetailTouchPoints.com: Seventy-one percent of consumers worry that an AI agent acting for them could give others the wrong idea about their personality, ethics, or values. How can retailers make sure shopping agents understand customer preferences correctly, and what could happen if they get it wrong?
Curtis: This finding shows that consumers are thinking about more than whether an AI agent will choose the wrong size or color. They also want to know whether it will understand and represent them correctly on a deeper level.
Retailers need to show customers which preferences an AI agent is using. Customers must be able to correct those preferences, make them more accurate, or choose to ignore them. If an agent keeps making choices that do not match who a customer is or what they care about, it can damage trust. As AI agents become a bigger part of shopping, representing customers correctly will be important for keeping their trust and business.
RetailTouchPoints.com: Only 33% of consumers would let an AI agent read their main email inbox, even if doing so saved them five hours of administrative work each week. What does this reluctance tell retailers that use personal data to offer personalized shopping experiences?
Curtis: Making life more convenient is not enough reason to ask for unlimited access to personal information.
Retailers should stop assuming that collecting more data automatically creates a better customer experience. AI personalization should only ask for the information needed for a specific purpose. Retailers should explain why they need it and let customers decide how long the permission will last. If customers feel they must give up access to too much personal information just to use an AI-powered service, retailers could damage the trust that personalization is supposed to build.
RetailTouchPoints.com: More than half of consumers say it is more important to know when AI is being used than to get the fastest result. Also, 20% would leave a website immediately if they thought it was using AI without telling them. How should retailers explain their use of AI without making shopping harder? Should they handle this differently in different markets?
Curtis: Being open about AI does not mean adding another page of small-print rules or showing a pop-up that interrupts every interaction. It can be as simple as clearly telling customers when they are talking to AI, what the AI is doing, and when it needs their approval before taking action. Consumers also told us that trustworthy AI should explain why it makes recommendations and offer a clear way to speak with a real person when needed.
Retailers should also pay attention to differences between regions. In the U.S., 63% of consumers say knowing when AI is being used is important. In Australia, that figure is 68%. In Indonesia, 62% of consumers place greater importance on speed and accuracy than on being told when AI is being used. The basic idea of being open about AI should stay the same, but retailers can change how they explain it to match what people in each market expect.
How Can Retailers Rebuild Trust While Asking for More Customer Data?
RetailTouchPoints.com: Only 12% of consumers fully trust organizations to manage their identity data, down from 17% last year. What can retailers do to rebuild trust while asking shoppers for more information to support AI-powered experiences?
Curtis: Retailers should not assume that consumers will automatically share more personal information in exchange for the convenience of AI. Every new request for access must come with a clear explanation of what customers will get in return and what the limits are. This starts with collecting only the information that is needed, explaining how it will be used, and giving customers meaningful control over their permissions.
As AI agents become a major part of online shopping, retailers also need to tell the difference between a customer and an AI agent acting for that customer. They need to know who or what is taking action, whether the agent is truly connected to the customer, and exactly what the customer has allowed it to do. Trust will come from giving consumers a better view of and more control over their identity and personal information, rather than simply asking them to believe that everything is working properly.
RetailTouchPoints.com: Three out of four consumers say government rules that protect personal data in AI applications are important. How should retailers prepare for new rules, and what can they do now to get ready?
Curtis: Retailers should not wait for government rules to tell them what the lowest acceptable level of protection is. Instead, they should build AI experiences that put consumers in control from the beginning. Businesses must be able to check that an AI agent is genuinely acting for a customer and make sure it stays within the permissions that customer has given it. They also need to keep clear records of what the agent does so they can show that they are responsible for its actions. Rules will continue to change, but retailers that build trust and customer control into their systems now will be better prepared for new requirements in the future.
