An AI-enabled app that makes purchasing recommendations, records your preferences and steers you towards the top items is really handy. But if users feel the app threatens their decision-making autonomy, they’ll steer clear of it – that’s the key finding of a recent article co-authored by NEOMA’s Mariyani Ahmad Husairi. At the same time, Ahmad Husairi and her fellow researcher are keen to point out that certain kinds of product don’t follow this general rule.  

Which would you rather trust: your cognitive skills or an app? Would you prefer to spend hours comparing different products or farm out the task to AI? Tap into your need for autonomy or “out-source” the grunt-work to technology? These are just some of the questions the researchers address in an article that draws on three studies carried out among 1,700 US consumers. The paper sheds light on our complex, ambiguous relationship with AI when we shop online.

AI is even in our fridges!

AI has already permeated our everyday lives. For example, it’s AI that selects the information about our favourite things for our Facebook or Instagram newsfeeds. The same technology generates recommendations based on our previous choices when we’re on Amazon, Spotify or Tinder. And it’s AI once more that automatically plays the next YouTube video, which more often than not is in sync with our interests.

Companies are developing AI for everyday shopping that predicts when consumers will run out of a given product, and which then triggers the next delivery in advance. And there’s talk of a near future where smart refrigerators will be able to place orders without any human intervention. That’s after they’ve checked out how fresh the food is in your fridge and planned your recipes for the week ahead!

In terms of the technology, AI is in a position to carry on ramping up. But for users, this means relinquishing their autonomy. Are consumers ready to give up trying to find the best products for themselves and stop deciding which ones to buy? Many researchers are sounding a note of scepticism, warning there’s a risk that AI that deprives us of these freedoms might be rejected.

The freedom to choose and make up your own mind

This is the hypothesis (which up to now had been developed in theoretical work) that the two authors have verified after carrying out three field studies.

In the first, participants used an app with four versions and four sub-groups to buy items of food. Two of the versions allowed the volunteers to search for the products themselves, with the final decision left to the user in the first case, and to AI in the second. With the other two models, participants were given a limited selection of products they could buy, with the decision once more left to the user or enabled by AI.

The results were clear-cut: participants were much more likely to adopt the app when they had the freedom to choose and make their own decisions.

AI: persona non grata even when the choice is complex

But do things change when users are faced with more complex choices? This was the question put to participants in the second study. Exactly the same app was used with its four different versions but the product descriptions – which had been concise in the earlier study – were now deliberately more convoluted. They featured some 20 or so characteristics, including smell, texture, cooking time, expiry date, nutritional values, origin and chemical composition.

It was found that we have limited attention spans and cognitive resources when it comes to processing complicated information. AI, on the other hand, is highly effective at performing the same tasks. So, did participants think it would be a good idea to piggyback on its processing power?

Nothing of the sort! They opted for the version that gave them the freedom to choose and come to their own decisions even if it made the search process more laborious.

Hobbies and leisure activities: AI is welcomed with open arms to help make decisions easier

The researchers selected a different type of product for the third experiment: running shoes. Our leisure activities are one of the ways we assert our sense of self, playing or doing things we identify strongly with. We put more time and effort into purchasing specialist equipment than we do into buying everyday supplies.

The researchers first assessed how keen participants were on running. They then used a running app connected to a site that sells running shoes. Participants were once more given four purchasing methods: free or limited choice, with the final decision taken either by the user or AI.

And this time, the results were different. Whereas participants with no great interest in running welcomed the ability to make up their own minds, avid runners plumped for the narrower range of choices – although they did prefer to retain control over the final decision.

How can we interpret this? It’s possible that AI becomes acceptable when it saves time and effort for the most highly-motivated athletes. Or, to put it another way: it would probably be better to treat yourself to an extra run than spend hours looking for what to wear on the internet.

Letting users make up their own minds

What are the key takeaways from these studies?

First: the technology needs to be kept on a tight rein so that it doesn’t scare users away. Why? Because we’re not ready to take on board AI that tells us: “You’re going to have a limited choice and I’m going to decide for you”. Plus, we want to keep a certain degree of autonomy. So, for example, the choices we’re given can’t be confined to the results generated by an algorithm. Users must have the power to look for products in sync with how they’re feeling that day or what’s happening in their lives or on the spur of the moment so they can still have the “freedom to do otherwise”.

The second recommendation, which is crucial for this perceived autonomy, is to leave the final purchasing decision to the consumer. And yet, this is already not the case with certain emerging business models. With shipping-then-shopping, for instance, a selection of AI-determined items is sent to customers, who then have to return anything they don’t want. Only the future will tell whether these deals, which are based on a “presumption of purchase” by the consumer, win us over or push us away.

Find out more

Mariyani Ahmad Husairi, Patricia Rossi, Delegation of purchasing tasks to AI: The role of perceived choice and decision autonomy, Decision Support Systems, 2024. DOI: https://doi.org/10.1016/j.dss.2023.114166