Dunn: Most retail AI efforts are doomed due to poor data

Edwina Dunn, one of the most highly respected data professionals in the business, has warned the retailers industry’s drive to implement artificial intelligence is likely to bomb due to the fact that most firms simply do not have the quality of data required to make it work.

Dunn has been at the forefront of data science for nearly three decades, having co-founded DunnHumby in 1989 with husband Clive Humby, and made her claim in an interview with The Retail Bulletin.

However, she does believe Tesco – which took a majority stake in DunnHumby in 2001 and bought the couple’s final 10% stake at the end of 2010 – does have an advantage.

While Dunn is hardly the first to raise fears over the quality of data needed to train AI systems, she argues that, despite the work undertaken by DunnHumby in delivering value from data, many companies still have woeful customer and business information. Too many firms continue to miss the point of what can be achieved and this places them in a weak position with the emergence of the technology, she argues.

Dunn explained: “All the AI players will rush their products to market and we will find retailers will be bitten as their data is no good. This will be a reflection of what has been the case for years. There’s not any intelligence [in the market].”

Even so, she reckons her former employer has the edge, adding: “I’d never bet against Tesco to deliver with AI.”

In fact, the supermarket giant is already exploring the use of generative AI within its Clubcard loyalty scheme, and how the next generation of the technology could be used to improve a customer’s shopping experience. It has also launched an AI-powered range curation tool, enabling improved tailoring of its store offer to local shopping habits.

Tesco CEO Ken Murphy told the FT Live Future of Retail conference in September last year that the company had already “radically improved” its supply chain efficiency, ability to forecast and how its stores picked on dotcom orders through the use of AI.

However, while AI seems to have everyone in the marketing industry in its grip, examples of how the tech is changing the sector still appear scarce. In fact, some observers believe AI is still not reliable enough to be trusted with many business decisions, with Apple’s recent woes just one example of how things can go badly wrong.

Earlier this week, the tech giant vowed to tweak its notification summary after a series of cock-ups, including a report that the suspect charged with the fatal shooting of the United Healthcare CEO Brian Thompson had shot himself. And, just last week, it crowned the teenage sensation Luke Littler the PDC World Darts Champion before the final had even taken place.

Dunn: Most retail AI efforts are doomed due to poor data

Be the first to comment on "Dunn: Most retail AI efforts are doomed due to poor data"

Leave a comment Cancel reply