Retail Pulse Report: Consumer Discontent with the AI Future is Growing
Consumers keep saying “I'm not sure this is a good idea” from the back seat. But they're not the ones driving.
Source: Adobe Stock. Prompt: “angry consumer”
This week’s topic will cover the question “What do consumers really want out of all of this agentic commerce stuff anyway?” It’s part one of two, because I only got halfway through and it’s already long. Part one, this part, looks at what consumers say they’re doing, and what they say about how they feel about it. Part two will look at how retailers are responding.
Remember when I talked about the Abilene Paradox? Consumers are like the mother-in-law in that story, except instead of waiting until the family gets back to say she never wanted to go in the first place, they’re sitting in the back seat on the way saying, “I’m really not sure this is a good idea.” Unfortunately, they are not the ones driving. And they know it.
Also, I’m not covering economic indicators this week – last week covered a lot and more coverage of July will really come out in the next week or two anyway.
Let’s jump in!
What Consumers Say They Do – And Won’t Do
We’re going to start with consumer behavior, in part because I found 4 big research reports in the last week that all try to get at whether/how consumers are adopting agentic commerce and how that is influencing their shopping behavior overall. As always, my grain of salt: consumers will tell you what they think they do, not what they actually do. The only way to get that is to observe them doing it at a time when they are not actively conscious that you are observing them. And also, with that caveat in mind, “what they think they do” is still interesting and valuable.
First up, a report by Commerce and Paypal looking at agentic AI shopping. Because these things matter more than ever with the speed of how things are moving, this is a 3,000 shopper study (1k each US, UK, and Australia), and it was fielded in March & April 2026. So a fairly recent take. The report I have only covers the US.
They found that 66% of US shoppers report using AI tools overall, and 28% specifically for shopping online. 67% say they are somewhat or very interested in an autonomous tool that shops for them, but 82% of respondents say the security protection has to be at least as good as, if not better, than regular payment methods.
As far as who would they trust to provide an agent for them, no one really got over 1/3 of respondents’ support. Tech companies were trusted by 34%, marketplaces by 32%, traditional retailers and payment providers tied at 31%, and platform LLM’s (OpenAI, Anthropic) came in at 17%. Which is funny when you consider how much Google and OpenAI are falling all over themselves to try to own the space.
Next up, an Ipsos survey into how AI is “reshaping the purchase journey” – almost 8,000 “AI-aware consumers” across 15 markets. It draws on multiple surveys fielded mostly through the first half of the year.
They found that 27% of AI-aware consumers use AI when shopping (primarily for product research). It’s tempting to say that this is very similar to the Commerce/PayPal survey, but Ipsos actually cut out people who said “what’s AI?” in the screening questions. It’s highly likely that most shoppers in the US, UK, and Australia are already aware of AI, so it’s probably not a huge mistake to take Commerce/PayPal’s 28% and Ipsos’s 27% and say “OK, I have something approaching triangulation here.” But again, caveats apply.
71% of respondents to the Ipsos survey believe that AI could be dangerous and needs more caution. Overall, 45% trust AI companies with their data, though that ranges to the extreme: 73% in India trust AI companies, only 25% in Canada. 57% of US respondents want to see the AI’s reasoning behind every product recommendation, no matter the value of the product in question. In general, though, respondents would be more willing to automate the purchase of high-frequency, low-consideration, and most importantly low price-point items.
There were other cuts by income and age – Gen Z is more likely to trust AI than Boomers, 48% to 33%. 53% of high-income earners trust vs 35% low income.
Next up, research from PYMNTS, looking into consumer attitudes and behavior towards AI discovery. This one is hot off the presses – just shy of 2,000 US adults surveyed in June 2026. Based on the results, PYMNTS estimates that 49.6 million American adults now begin product research with AI, and 39 million have completely abandoned other search methods. PYMNTS can make this claim presumably because they feel they have enough of a representative sample that they can make these extrapolations. The millions of user numbers definitely sound more impressive than “19% of respondents.”
Of those who have used AI to make a purchase, 43% say they found a better price, 27% bought a different brand than they were expecting, and 26% bought a different product.
Based on these numbers, PYMNTS estimates that “AI-assisted spending” has hit around $74.1 billion in the US, and $15.1 billion was directed to a new retailer by AI. They also claim that Google search has lost 9.5 million shopping research starts to platform LLM’s (this is separate from those who have moved to use Google’s own AI features), and that review sites have lost 1.2 million research starts.
I get what they’re trying to do – take %’s and translate to total impact. Let’s say I accept that their methodology for getting a representative population is good, it’s still problematic to make these projections and claims based off of people answering questions vs. watching what they actually do.
Having said that, even if the numbers are half right, it is still a sea-change in consumer behavior.
And finally, last research report, this one from The Harris Poll, but more focused on consumers’ attitudes towards AI and trust. They don’t say when the survey was fielded, only that the report was published in June 2026. They surveyed 3,222 adults across the US, UK, Brazil, and India, and focused in on beauty & wellness, personal care, and household goods. 72% of respondents are comfortable with AI helping them shop – as long as they get to keep authority over the final decision. The use cases they are most comfortable with are comparing prices (74%), checking product-claim accuracy (71%), discovering new/emerging brands (71%), and summarizing reviews (71%).
There are some caveats, though. 76% would lose trust in a retailer whose AI recommended sponsored products without disclosure, and 78% assume brands will try to pay their way into AI recommendations.
72% worry AI will limit rather than expand what they see, which is interesting when you consider that the leading use case for AI shopping is a research start. In the abstract, the idea is that you give specific details about what you want – better than you can use to refine a facet-based search on a retailer’s site, for example – and you’ll get a narrower, more focused set of results. But I think that plays to one of the Ipsos survey results, that 57% of US shoppers want to see the reasoning behind every product recommendation, even the ones that are low-consideration / low-value. One might sum that up as “trust, but verify.”
Another interesting set of results: 73% agree shopping would feel “less fun” if AI did too much of the choosing. I see this in my own shopping. I am interested in the reasoning that an AI gives me as it makes recommendations, as much to see if it’s detecting a pattern or preference that I haven’t seen in myself or explicitly given a name, just as much to make sure it didn’t glom on to something weird that I didn’t intend when understanding what I am looking for.
It sounds like consumers in general are also struggling with that balance. 71% of respondents said they’re comfortable using AI to find new brands. At the same time, 61% say they’d be less likely to ever notice a brand AI doesn’t recommend.
I think one take-away for retailers and brands for their own on-site chatbots is to possibly over-index on explaining the why behind a recommendation – and letting consumers adjust the dials on those assumptions. The retailer gets to learn a lot, and the consumer gets to build trust that the recommendations they get are tuned specifically to them.
How Consumers Feel About What They’re Doing
If you’ve read more than one post from me, you will know I am a fan of Cory Doctorow’s term “enshittification.” He’s developed it really more into a Theory of Enshittification and even written a book about it. The very short take on what it is: the inevitable decline in the quality of a product or service in the pursuit of extracting more profit.
It’s especially evident in marketplace plays, where a company will provide a great service in order to attract a large base of customers, then uses that customer concentration to attract sellers, and once sellers are locked in and dependent on the marketplace for their own customer base, and consumers are habituated in coming to the marketplace for everything, then the marketplace can start extracting value from both sides because they don’t have as many options on where else to go.
So full points to Matt Klein and ZINE for commissioning a survey (via YouGov) to find out just how much consumers are aware of – and hate – enshittification. He surveyed 1,000 US consumers, I don’t have a date noted but the article published early August.
You want to hear jaded? 62% of consumers surveyed believe most companies prioritize growth over customer experience. Only 2% believe the opposite. 81% of respondents say they have taken an action against a brand in the past year after a product or service degraded – 45% canceled a subscription, 38% say they left the brand entirely, 35% delayed a purchase, and 12% traded up to a pricier alternative just to get back to the quality they wanted.
The generational splits are really interesting, especially when you consider that Gen Z is overall more trusting of AI for shopping. 53% of Gen Z say they would leave a brand over paywalled features vs. 31% of Boomers, and 40% would leave over more ads/upsells vs. 24% of Boomers. On the flip side, 54% of Boomers would leave a brand over unreachable customer service (vs. 27% of Gen Z) and shout out to Gen X for 40% saying they would leave a brand over “feeling like a number” vs. 21% of Gen Z. In general, the intervening generations fall along the slope of the line you’d draw between the two bookend generations.
The take-away here is, Gen Z has a higher tendency to punish digital extraction of value while Boomers and Gen X tend to punish “dehumanization.”
I mean, some of this is wishful thinking on the part of consumers. It’s hard to sustain anger over the ease of convenience. I once tried to not shop from Amazon for a month. It was both incredibly time consuming, and ultimately more expensive. I might pay more to save time, and I might spend more time to save some money, but I’m certainly not going to pay more AND spend more time at it.
But it does still hint at a real fragility. If anything better came along and promised to take my whole recurring Amazon purchase history and re-source it every time, ultimately executing across Amazon as well as any other retailers to get me the best deals, reliably delivered, I would probably take a serious look at it. And guess what is getting closer and closer to promising to do that? Yeah, agentic commerce.
That fragility also comes with a cost. Klein put some example numbers against it, with Klarna’s rollback of AI-driven customer service as one example, or Amazon’s $2.5 billion settlement with the FTC over how difficult it made its Prime cancellation promise.
And Bain points out that a +5% improvement in retention lifts profits 25-95%. This kind of number has been floating around retail since basically forever. AI hasn’t done anything to change that math (at least not yet). But retailers and brands still can’t bring themselves to give up the sugar-hit of customer acquisition in favor of the protein-rich engaged and loyal customer base. It’s all short-term thinking, which is exactly what enshittification ultimately is as well: if I cut just this little bit here, then I can drop that savings straight to the bottom line and no one will really notice. As Klein points out, enshittification doesn’t show up in dashboards.
Another way to frame this: death by a thousand cuts. You’re fine and it’s all good and it’s still fine, until suddenly it’s not. It’s just that in the internet age, we haven’t seen a big brand die this way yet. Lots of other things have killed brands, but not this, heavy emphasis on YET.
Side note, Klein refers to the National Rage Survey, which I think I’ve run across but also forgot until he pointed it out. The fact that this exists is both awesome (in the what a great idea sense of “awesome”) and depressing.
Sharon Edelson digs into this idea of customer frustration with AI, bringing even more examples of research and anecdotes to the table in her Forbes piece on the topic. She highlights Gartner research, along with some quotes I like from a couple of people in the industry:
Gartner: only 11% of US consumers are willing to let AI make purchase decisions even for low-stakes categories (personal care, household supplies). 62% say AI shopping recommendations are a waste of time and 49% agree generative AI has worsened content quality
Rebecca Brooks (CEO, Alter Agents), on the “trust recession”: “Consumers have a lot of distrust in all sorts of institutions and brands... Consumers are seeing the race to AI without thinking about the consequences.”
Molly Moore (President/COO, Liveops), on the “Resolution Gap” study her company published: nearly 60% say bot-to-human handoffs break down badly enough to restart from scratch. “Consumers aren’t rejecting AI so much as getting increasingly irritated with experiences like shopping that make more work for them.”
Sonia Lapinsky (AlixPartners): “It’s up to brands and retailers to get smarter about governance... There is a little bit of a fear that these things are coming out of control.”
Gartner projects AI platforms – those platform LLM’s that only 17% of consumers said they’d trust to deliver commerce agents in the Commerce/PayPal study – will drive over $20B in US retail eCommerce spend in 2026. They also say that 90% of B2B purchases will run through AI agents by 2028 - $15 trillion running machine-to-machine. (Gartner likes to come up with really outrageous numbers, sometimes I think just to get quoted – all publicity is good publicity even if someone is making fun of your position – and clearly it works because I just quoted them.)
That’s OK. PYMNTS says nearly $50B of retail eCommerce spend is already AI-assisted today. And Coresight says $943B in retail sales will be “mediated by autonomous AI agentic commerce” by 2030. So suddenly Gartner doesn’t seem quite so reckless.
But it’s no wonder that consumers’ anxiety about AI is turning into outright anger. This begins to feel more like getting railroaded than agreeing to go along for the ride.
What Have We Learned This Week?
There’s more, there’s so much more, to say on this topic.
For now, let’s focus on the first half of the message: consumers are using AI, but they don’t love it. I think they’re starting to get how this enshittification thing works, and they’re digging their heels in early. Just like they’re hitting the NIMBY brakes on data centers. They’re trying AI, they see that it kinda works and it does maybe save them some time and maybe save them some money and maybe get them a better result, but they can also see how all that can be squeezed to benefit someone else at their expense once they’re hooked.
Retailers definitely sense the potential to be squeezed here, because they’re fighting right now to keep the front door to consumer intent. Ironically, if they genuinely cared about their customers and their customer experience (instead of thinking how to shave just a little bit more off to preserve or grow margin), they wouldn’t have to worry about the front door because the customers would already be inside the house.
But I also get that economic conditions mean you can’t be complacent about the customers you already have. If things get too expensive, or consumers just plain feel like they have less disposable income to spend, then your products will get handed off to an AI to find a cheaper replacement, sometimes even when the experience and loyalty is already great.
If consumers are in the back seat of the car on the way to Abilene saying “I don’t really want to go there” then who is driving the car? The platform LLM’s are. Retailers just happen to be in the front seat, and I think even they are not sure they want to go along for this ride.
But we’ll dive into that more next time.
Until then!
Nikki
I write this weekly article as much to force myself to make sense of what I read on a daily basis as to say anything about it publicly. I will continue to do so - for free - for as long as I can sustain it.
If you like what you’ve read and you haven’t subscribed, please do! It helps me know I’m saying interesting things.
If you’re already subscribed but you can think of someone who would enjoy reading this, please share the article.
Or better yet, share the whole publication and let them get in on the weekly fun!
Thanks for reading, and thanks for your feedback and support!