Who is the Trolley Working For?

Pt 2 of 3: From Twin Tub to Trolley Tech

In Part 1, I explored what technology might give back to us. Now I want to turn the trolley around and ask a more uncomfortable question: who is it actually working for?

There is an exciting version of the future in which technology genuinely understands what I want. Imagine an AI-powered supermarket trolley that knows the dietary requirements I have given it, the foods I enjoy, the budget I have set and perhaps some of the health goals I have chosen for myself. Instead of navigating thousands of products and hundreds of tiny decisions alone, I have something helping me. I like that possibility, except the technology that knows enough about me to help me make my choices also potentially knows enough about me to influence them.

How free are our choices now?

Long before artificial intelligence arrived, supermarkets understood something about human behaviour. Where something is positioned matters. What appears at the end of an aisle matters. Price, packaging and promotions matter. So does that wonderfully catchy little acronym, BOGOF — buy one, get one free — or its close relative, 3 for 2.

I know the effect these things can have because I have experienced it myself. I have gone into a supermarket intending to buy one of something and bought two or three because the deal made buying more appear to be the sensible choice. I have bought something I had not intended to buy because the offer seemed too good to miss. There is another uncomfortable part too. I know there have been times when having more of something in the house has meant that I have consumed it more quickly.

Did I save money? Possibly not. Were they my choices? Yes. Nobody forced me to pick anything up or physically put those products into my trolley. Yet that does not stop me being curious about what influenced the choice I made. Would I have bought three if one had been the better deal? Would I have bought it at all if it had not been positioned prominently? Would I have wanted it quite as much without the large sign telling me how much money I was apparently saving?

I cannot answer those questions for anybody else, but I can ask whether you recognise any of them.

There is evidence that this matters beyond my own experience. UK Government evidence used in developing restrictions on promotions of foods high in fat, sugar and salt found that volume promotions encourage people to purchase more than they otherwise would and that the additional food purchased does not simply sit untouched in the cupboard. The evidence suggested that increased purchasing can also result in increased consumption.

This contributed to restrictions in England on volume-price promotions such as BOGOF and 3-for-2 offers on specified foods high in fat, salt or sugar. Restrictions have also been introduced around prominent locations such as checkouts, store entrances and aisle ends. We have therefore already recognised something important: the environment in which we make a food choice can influence the choice we make.

Then personalisation enters the room

This is where things become even more interesting for me. Sainsbury's already offers Your Nectar Prices, providing customers with personalised discounts based on their shopping history. The more information the system has about what somebody buys, the more personalised those offers can become.

On one level, I love this. Who does not enjoy discovering that something they regularly buy happens to be cheaper this week? It feels useful, relevant and personal. There is something psychologically different about an offer and my offer. It can almost begin to feel less like Sainsbury's and more like Sarah's-bury's.

Except I have an underlying discomfort with that too. The better a retailer understands what I buy, the better it potentially understands what might persuade me to buy. The distinction between personalisation that serves me and personalisation that becomes increasingly sophisticated marketing may not always be obvious from my side of the trolley.

That does not make personalised offers inherently bad. They can genuinely save us money. It simply makes me want to remain conscious of the fact that something can feel wonderfully personal while also being commercially valuable to the organisation providing it.

Now make the trolley intelligent

Imagine taking that knowledge and putting it into the AI trolley. It knows what I usually buy, what I have bought before, what I respond to, what I reject and perhaps eventually what I am trying to change. That could be extraordinary.

I could tell it that I want to eat differently, spend less or follow particular dietary requirements. It could reduce some of the friction involved in doing what I have decided I want to do. Except we need to be careful about what we mean by help.

Imagine I pick something up and the trolley effectively tells me, No. You said you were not going to buy that. I am not sure that is freedom either.

There is a difference between support and control

Human beings have an interesting relationship with being told what to do. Sometimes we comply. We can be the “good girl” for quite a long time, following the rules, doing what we believe we should do and congratulating ourselves on getting it right. Until sometimes we do not.

At some point external control can become exhausting. We snap. We rebel. We do the very thing we were trying not to do and sometimes do rather more of it precisely because we are reclaiming the freedom to choose.

Psychology has a term relevant to this: psychological reactance. It describes the response that can occur when we perceive our freedom to choose as being threatened. One possible reaction is to become motivated to restore that freedom.

That makes helpful technology considerably more complicated than simply making it clever enough to tell us what to do. Perhaps the helpful trolley does not say, You cannot have that. Perhaps it says, You told me this was important to you. Would you like some alternatives?

One approach controls; the other gives the decision back to me. Suddenly this is not really a conversation about supermarket trolleys at all.

Perhaps this explains something about our response to AI

I wonder whether some of the resistance we see around AI contains an element of this too. Some people are fascinated by it and excited by what it can do. Others do not trust it, do not want to use it or want nothing to do with it. There are perfectly rational reasons for concern. Privacy, misinformation, employment, ownership, power and environmental impact all deserve serious discussion. There is also something very human about not wanting change imposed upon us.

Think about something as ordinary as a supermarket self-checkout. Some people avoid them because they do not want checkout staff to lose their jobs. Others dislike the machines because they find them frustrating. Some simply prefer handing their shopping to another human being and having a conversation.

That matters because efficiency is not the only thing of value in a supermarket. The person on the checkout may be one of only a handful of people somebody speaks to that day. The familiar member of staff who notices someone has not been in for a while, the tiny conversation in the queue, the person who helps when something will not scan — none of those things appear on a spreadsheet measuring checkout speed. They still have value.

Community is not inefficiency. Human interaction is not wasted time.

That feels particularly important when we talk about AI. The question cannot simply be Can technology do this? It also needs to be What would disappear if it did?

We are happier about some machines than others

There is an interesting contrast here. Many of us seem far more comfortable buying technology for our homes that removes work. Robot vacuum cleaners, dishwashers and countless other domestic gadgets have been enthusiastically welcomed because they promise to make life easier.

Yet perhaps even there it is worth asking who told us we needed them. What problem does this gadget actually solve? What value does it bring to my life? Is it genuinely freeing my time, or have I simply acquired another device to buy, maintain, charge, update and eventually replace?

More importantly, if it does give me twenty minutes back, what happens to those twenty minutes? That question takes me straight back to the twin tub.

Technology cannot tell me what enough looks like

This is where the conversation starts to move into something much closer to my own work. If I do not know what I want, there are plenty of people, businesses, algorithms and machines increasingly capable of suggesting answers: buy this, achieve that, work faster, produce more, upgrade, optimise, save time and then spend the time you saved becoming even more productive.

Perhaps our own voice becomes more important as technology becomes more capable, not less. I need to know what matters to me before I can decide whether a piece of technology serves it. I need to understand what health means to me before I ask an AI trolley to support it. I need to know what I value about community before I decide whether removing human interaction represents progress. I need some idea of what I want my future to look like before I can decide which efficiencies actually move me towards it.

Otherwise I risk becoming incredibly efficient at travelling towards a destination I never consciously chose.

That leaves me with a much bigger question than who is controlling the supermarket trolley: why do we do what we do — and why do we not always do what we wanted to do?

Perhaps it is not enough simply to have choices. Perhaps autonomy requires us to be able to hear our own voice amongst all the others competing for our attention.

That is where I am going next.

Part 3: Whose Voice Is Choosing the Destination?