Instacart’s manipulative AI jacks up the cost of groceries

A study from Groundwork Collaborative, Consumer Reports,and More Perfect Union found that Instacart charges shoppers more for the same items, depending on how likely they think they are to spend

Ever since New York City mayor-elect Zohran Mamdani rose to victory on a platform of affordability, it’s become the buzzword in the public discourse. Even Donald Trump, recognizing the concept’s power in affecting voters’ mindsets, is making a point to wrest control of the narrative, calling the affordability crisis a Democratic “hoax.”

As Democratic politicians lean into affordability crisis messaging and Trump tries to deflect, tech companies are doing something else altogether: quietly taking advantage of it. With the understanding that some consumers might be more numb to price or less sensitive to spending more on brands or items they like, tech companies are developing pricing algorithms to capitalize.

Last week, a report from several consumer rights non-profits and investigative news outlets found that on-demand grocery app Instacart is leveraging what they call “algorithmic pricing” or “smart rounding” to test just how much more grocery shoppers will spend on items. The study enlisted volunteer shoppers to purchase identical products, and found that Instacart offered most shoppers the same basket of groceries at 5 different prices, ranging from $114.34 to $123.93. From the report:

The price of the same basket of food at a Seattle-area Safeway on the Instacart platform, for example, ranged from $114.34 to $123.93—roughly a $10 difference. And only 8 percent of those shoppers got the lowest basket total.

Since the dawn of digital products, designers have been playing with features that subtly induce users to take the action the product wants them to take—namely one where the user is spending more money or more time on the app. But a lot of that user interface design was a result of broad testing to understand user psychology: for instance, is someone more drawn to click on a certain color, or motivated by an animation that makes them feel they are running out of time to buy something. These practices are known as “dark patterns,” or deceptive design patterns.

But here we are dealing with something far more sophisticated, something “pernicious” that the authors of the report refer to as “surveillance pricing.” Not only are tech companies formatting user displays to appeal to what our brains might engage with–but they’re mining our personalized data to potentially get us to spend more.

With “surveillance pricing,” a company like Instacart can tap into a trove of someone’s personal characteristics, behaviors, and past shopping history to set individualized prices. And the average price variations revealed by the study could cost a four-member household about $1,200 per year.

While Instacart, in response to the study, called the price variations “negligible,” the variations are actually not insignificant to individual people, as Mamdani’s win shows. In very human terms: as I was checking out at a Manhattan grocery store yesterday, I overheard a man at the register next to me say to the cashier, “I thought this was $3.99, not $5.99.”

So yes, while Instacart may call these price differences “negligible,” actual people are going into stores and questioning an additional $2.o0. (In New York, when you think about the fact that $2.00 is most of a subway fare to get someone to and from work, you start to wonder whether it actually is so “negligible.”)

Without a human cashier to look straight in the eyes and ask such a question about $2.00 to, you may be handing over your agency to these platforms like Instacart, engaging with an interface that is extractive and almost always further alienates people. At the time, it might seem like a platform like Instacart is making your life a lot easier and saving you time. But that assumption is also giving the platform more control to set prices–and you may be losing $2.00 each time because Instacart’s algorithm has deduced that you have less price sensitivity because you really love Wheat Thins or Skippy’s peanut butter.

Platforms like Instacart may now have just enough personalized data on someone like the man I encountered at the grocery store to know that he is extremely price-sensitive, and that he would contest $3.69 peanut butter over $2.99 peanut butter. So they may offer him a lower price, while charging someone else with less price sensitivity more.

Either way: the consumer is the guinea pig, and Instacart’s retail partners and shareholders have good numbers to share. And the affordability crisis remains.

Anyway, here’s what we’re reading…

There are some smart billionaires. So why are they so clueless about how the public feels about them? Perhaps it’s their increasing isolation: The Billionaires Have Gone Full Louis XV

Speaking of isolation, writer Rebecca Ackermann makes the case for taking public transit instead of autonomous Waymos: “As a parent, I’m also deeply concerned about the message that kids are getting that a vehicle without a person in it is safer than one with. While human interactions do come with risks, the approach to keeping our kids and ourselves safe should be education, conversation, and community, not tidy isolation.”

A few week ago we wrote about scammers paying Meta to list their ads. Reuters is out with a new deep dive into just how much money Meta was knowingly making from Chinese scammers and affiliates: Meta tolerates rampant ad fraud from China to safeguard billions in revenue

There’s a new crop of publications catering to techno-optimists: “The typical reader is a young techie, but the kind who added on a humanities minor in college and still writes poetry on the side.”

Bernie Sanders is pushing for a moratorium on the construction of data centers that are “powering the unregulated sprint to develop & deploy AI.”