AI in Retail: The big debate

Welcome back to dunnhumby's?The Science of Shopping?newsletter where we look at the trends shaping the future of retail, and share the top insights that matter most to retailers and CPGs. In this month's edition, Sahira van de Wouw, PhD dives into the world of artificial intelligence (AI) and what it means for today's shoppers .

One of 2023’s hottest topics also happens to be one of the great issues of our time: artificial intelligence (AI). From the short-term impact on jobs to the long-term outlook for humanity as a whole, people everywhere are beginning to ask what the rise of AI might ultimately mean for our species and the planet.

Overall trust is low, with older shoppers being particularly apprehensive

Public trust is absolutely fundamental to AI’s success. The more reliant we become on computers, the more important it is for people to have confidence in them, particularly in terms of AI’s decision-making capabilities. Indeed,?the ethics of machine-assisted decisioning ?was an issue that we covered all the way back in 2020, and the importance of moral stewardship in that area has only grown since.

From the results of this latest Shopper Thoughts study, a great deal of work remains to be done here. Well over a third (39%) of those that we surveyed said that they have no trust whatsoever in AI, and even amongst those who do, the majority (47%) are only “somewhat” confident in it. At the other end of the spectrum, only a miniscule – yet undeniably optimistic – number say that they are “completely” trusting of AI (2%).

Concern is particularly acute amongst older shoppers; more than half (53%) of all respondents aged 65 and above said that they have zero trust in AI.

Clear awareness of the threats, but opportunities recognised too

Regardless of whether they trust it or not, the primary fear amongst our respondents when it comes to AI is that of loss: they worry that AI will either rob them of the opportunity to interact with real people, or that it will inevitably result in redundancies and layoffs. A common response was also to mull whether AI would threaten the entire existence of humanity, with many respondents referring to the dystopia seen in the Terminator film franchise.

Given the chance to share their specific views on this subject, many respondents told us that “people rely too much on computers” and that “we need more human interaction”. Others took that line of thinking one step further, noting that they could “see AI replacing some jobs” and even that it represents “a massive threat to peoples' lives and employment”.

Those concerns aside, we do also see clear recognition of the additive impact that AI can have. Respondents share positive sentiment around a number of different issues, ranging from business processes to daily tasks. Key themes include efficiency (“a useful tool to speed things up”), repetitive tasks (“AI will make them simpler and quicker to do”), and information management (“it’s a sophisticated search engine”).

Of all the opportunities identified by our panel, however, AI’s potential impact on health and wellbeing was the one that drew the widest support. Amongst those respondents who agreed that it offered at least some benefit, 42% cited healthcare as an area in which AI could make a clear positive impact. The focus here tended to be on research and medicine, though applications within hospitals and other care facilities were also called out.

AI-generated imagery can prompt a negative response amongst shoppers

Generative AI – that which focuses on the creation of imagery, text, video, and audio – has also generated some of the fiercest reaction so far. One major point of contention behind the?SAG-AFTRA strikes ?in the US is that TV and film studios may attempt to use that technology to recreate actors’ likenesses without their permission. This follows on the heels of a similar motion by writers, who are seeking reassurances that they will not be replaced by AI.

If the use of generative AI in brand marketing materials might feel like safer territory, then it’s worth bearing in mind that many shoppers already have a keen eye on what they’re being sold – and just how realistic of a representation it really is.

Presented with AI-generated images of customers, for example, 19% said that they’d be less likely to purchase from a brand that used them, 15% said that they’d find the brand less trustworthy, and 13% that they’d see it as being less creative. Women typically have a stronger reaction here, with more than a quarter (27%) stating that they’d be less likely to buy from a brand that used generative AI to recreate customer lifestyle shots.

Interestingly, this reaction becomes stronger still when we turn to the subject of artificially generated product imagery. Here, the number of respondents who say that they would find a brand less trustworthy rises to 22%, suggesting that customers prefer a “true” representation of the product they’re buying – or, at the very least, one that was posed and photographed by a human rather than created by a machine.

Whether for inspiration or illustration, the message is clear: brands need to tread carefully, and use real images rather than computer generated images wherever possible.

Chatbots, no; ChatGPT… perhaps

Continuing the theme of human interaction referenced above, chatbots drew a generally negative reputation from our respondents. Fully two-thirds (66%) of respondents cited their dislike for them, with some even going so far as to stress that they “hate” automated communications. Specific use cases do seem to attract less ire, however; more than a third (37%) saw value in using chatbots specifically for survey purposes, for example.

One additional – and unprompted – learning here surrounds ChatGPT. On the subject of chatbots, some respondents took the chance to praise the functionality of OpenAI’s platform and its value as a learning tool. “It’s fab to get information straight away,” noted one of our panellists, suggesting that the integration of?Large Language Models ?(LLMs) into chatbots might help to assuage some of the current doubts.