Can artificial intelligence in retail ever be too smart?

The ultimate aim of artificial intelligence in retail isn’t necessarily friction-free purchases.

You’d be forgiven for thinking this was the case…but it isn’t necessarily true.

Retail experts speaking at the latest Retail Connections event in London explained that customer needs are far more complex, and in many cases consumers actually need to experience a manageable level of effort and stress during the product search stage to make the buying process satisfying.

Sliding scale of needs

The Retail Connections panellists described a sliding scale in which consumers want to buy everyday low-price products with minimal or no effort, while they need to spend time and effort selecting big-ticket purchases – even if AI can carry out the search in its entirety based on previous purchase behaviour.

Paul Wilkinson, Head of Tech Research at Tesco Labs, told the Retail Connections audience that AI is already advanced enough to learn consumer behaviour and automate purchases.

He revealed that in his own home he has developed a system in which an electricity smart meter counts the number of times his dishwasher is used and then, thanks to AI, an order is automatically placed for dishwasher tablets on Tesco’s website. His system has been running perfectly for 18 months with no human intervention.

Automating routine purchases

Paul’s wider argument is that soon it will be possible for all consumers to make routine domestic purchases automatically, once an AI platform has learned their purchasing patterns.

While AI has the potential to learn and predict consumer behaviour to the point at which virtual digital assistants will tell customers when they need to buy items or even buy them automatically, it remains to be seen if consumers are comfortable with this inverted relationship.

Anusha Couttigane, Senior Analyst of Kantar Retail, told the Retail Connections audience that organisations shouldn’t confuse past buying patterns with personal taste, warning that the impact of AI on the fashion retail sector may not be as great as grocery.

Multiple fashion variables

“There are so many variables at play when a consumer makes a fashion purchase, past behaviour isn’t always a good indicator what a consumer will want in the future.

“I really can’t see a future world in which AI will be able to pick an outfit for a customer with no customer input.

“AI will be great at making suggestions along the lines of ‘you bought this so you might like that’, but it will be a long time before they can replace a personal shopper, for example, who can say whether the colour or cut of a particular outfit suits a customer.”

This was a theme picked up and expanded by Katharina Wittgens, Managing Director of Innovationbubble.

Deep-seated human needs

Katharina, a retail psychology expert, explained that AI needs to be responsive to deep-seated human needs if it is to be successful.

She said: “Humans actually need to invest a level of effort if they are to derive value from a purchase.

“No matter how effective AI is at making a really good purchasing decision, we have a deep-seated human need to reassure ourselves that the product or service is right for us. We do this by going through and weighing up the options available.

“Some consumers may need to invest more time and effort than others to satisfy this need, but the need is still real.”

Watch out for blog posts giving further insights from our event Artificial Intelligence: the new game changer in retail

Join Retail Connections

Register here

Related insights


In a tough retail market, no one is immune


Rising to the challenge of international returns

Innovation trends

Retail returns: turning a nightmare into an opportunity

eTail Fulfilment & Returns 2018: From click to customer 


Your cutting edge may be someone else’s everyday tech


Stop listening to people who agree with you

Join Retail Connections

Get the latest industry views and exclusive member offers sent direct to your mailbox.