News and views for the UB community
Q&A
Will AI spell an end to shopping?
In a new paper, School of Law Professor Mark Bartholomew — pictured near the defunct Boulevard Mall near UB’s North Campus — discusses how AI shopping agents could impact marketplace decisions previously made by humans. Photo: Douglas Levere
By CHARLES ANZALONE
Published October 2, 2026
It’s one of the most routine things people do. So routine, that few have stopped to ask what happens when a machine starts doing it for them.
Shopping.
In a new William and Mary Law Review paper titled “The End of Shopping,” UB legal scholar Mark Bartholomew forecasts a world where artificial intelligence “shopping agents” take over marketplace decisions previously made by people.
“Deployed by major platforms like Google, Amazon and Walmart, AI systems are evolving from passive advisors to autonomous decision-makers capable of opening accounts, canceling subscriptions and finalizing purchases without human oversight,” the paper states. “Though AI shopping agents have the potential to improve consumer satisfaction and marketplace competition, shopping agents come with attendant risks.”
Bartholomew is vice dean for research and faculty development and a professor at the UB School of Law. His co-author is Samuel I. Becher, professor of law at the City University of Hong Kong School of Law.
They warn that without updated legal measures, this shift toward automated commerce “could quietly erode the economic, psychological and social benefits that people receive from shopping on their own terms.”
Bartholomew, an expert on the intersection of AI and the law, talked with UBNow about what he calls the rise of AI agents as a “critical challenge for the law and society, rather than a mere evolution in retail convenience.”
How are these AI shopping agents intruding in the lives of everyday shoppers? How will we know they have arrived in our spending landscape?
Every major player in artificial intelligence has already rolled out shopping agents, promising to automate much of the product research and selection involved in shopping. Sometimes this is advertised as a key selling point, as with Meta’s recent launch of its Muse AI agent. Other deployments of this technology are more subtle, with retailers using AI to improve their recommendation engines without full consumer awareness.
Messing with our shopping habits seems like an unforgivable violation to numerous people. What are the risks of this trend toward machines and algorithms making shopping decisions for us?
First, I want to note that there could be a lot of benefits to using this technology. AI agents can canvas so many more retail listings than a human being can possibly look at, potentially unlocking more variety or better deals.
In addition, these agents won’t be influenced by the irrational things, like celebrity endorsements or prices ending in 99 cents, that sway human shoppers. And AI could remove a lot of the hassle involved in things like switching internet companies or insurance providers that can currently cause us to stick with suboptimal services.
But the risk is that the shift to automation could replace the good parts of traditional shopping, too. For example, gift giving is a big part of shopping. A good gift depends, at least in part, on the recipient perceiving the gift giver’s effort and understanding of their wants and needs. Now imagine an automated system that not only remembers your anniversary for you, but decides and orders what your partner most wants for their anniversary without any input from you.
Likewise, shopping involves not just finding the best deal but seeking out certain products and brands to express who we are. What does it mean when AI starts to choose those commercial expressions of identity for us?
Can you continue along these lines and explain why consumers have good reason to be cautious and why shopping agents may favor particular businesses?
The most important thing in setting this new technology on the right path is making sure that it aligns with the shopper’s best interests. Users won’t be able to tell if their agent is actually recommending the best deal available. If, behind the scenes, an AI platform prioritized products that it made itself or that it received a commission from retailers for, that would drain this new technology of its promise.
Already, there have been studies showing that AI-assisted shopping increases the amount and expense of purchases someone will make. We want to make sure that shopping agents don’t become just another online feature that we worry might be scamming us.
Is this AI shopping model inevitable, or can legislation and personal awareness reduce the psychological and financial costs of AI shopping agents?
We propose a series of legal guardrails to ensure that AI shopping actually serves the best interests of human clients. It’s too hard for someone to figure out on their own whether an AI recommendation is truly trustworthy or not.
Better instead to mandate that shopping agents be audited by trusted third parties that can certify whether they are doing a good job or not. And we also recommend that there be design options — like a simple off button — to allow users to rely on old-fashioned human market research when they want to.
What got you interested in this topic?
In other work, I’ve explored the growth of advertising in this country, infiltrating places like public schools and national parks. I think too much commercialism in our lives is bad, but for this project I wanted to explore the good things about our commercial choices.
It turns out there’s a lot of fascinating research that has been done on shopping’s psychological and social benefits. “Retail therapy” is a real thing. This project was a way to utilize that research to help retain the best parts of shopping while still recognizing the value in this new use of AI.