AI fear and AI ethics: Decathlon explores the knots

With a steady drip of artificial intelligence experts warning about the acceleration of the technology (many while continuing to develop it), some discussion of practical AI ethics from practitioners is also now emerging.

Why it matters

What’s often missing in the AI conversation is how users of generative AI systems are thinking about the principles they want to embed in their own uses of the technology. Effectively, it’s time to lay out some rules that you would be prepared to defend later on should your brand come under scrutiny.

The view from Decathlon

Sporting goods giant Decathlon, headquartered in France, has been experimenting with the technology for its product development, according to chief digital officer Jérôme Dubreuil, who was speaking at this week’s Shoptalk Europe.

For the retailer, which designs as many as 90% of the products it sells in-house, the technology offers opportunities to speed up the iteration of product design, to better understand its vast trans-European user base, and to improve demand forecasting.

In the absence of regulation, Dubreuil says, ethics take over – and given how fast the technology is moving ahead of regulation, ethics are more important than ever.

For Dubreuil, this translates to three things that are an extension of data ethics:

- “The first one is auditability;

- “The second one is observability;

- “And the third one is explainability.”

“As we inject AI in so many of our processes, so many of our customer experiences,” he adds, “making our users comfortable with what is happening inside the box and giving our users choice is really critical.”

It’s a constant task to establish guidelines about what should and shouldn’t be done, and to keep as clear a record as possible – which is an individual company’s choice.

But he compares the emerging regulation, such as the EU’s proposed AI Act, to GDPR. Prior to that you could care a lot or you could care a little, doing only the basics, until you couldn’t. Then, ultimately, “there was no choice”.

Eventually, firms will also have to consider the effect AI systems may have on human labour. Dubreuil sounds a positive note: “We are not here to replace people, we’re here to augment.” But how this plays out in reality remains difficult for anyone to know.

In context

Since November last year, AI has been not only a concern for marketers, but for practitioners of many other disciplines across industries.

Recent interventions and comments from former Googlers like Geoffrey Hinton and Mustafa Suleyman could be read as attempts to inject a more thoughtful tone to the conversation, while repeated warnings from OpenAI’s own CEO Sam Altman have more of the feel of a marketing tactic; the sense of danger can inspire a sense of urgency along with general interest.

IBM last week said it was pausing hiring on jobs that could be done by AI, before unveiling its own system to help companies integrate AI capabilities into their product stacks, which definitely had the feel of ‘if we can do it, so can you’.

Reported by SPT. Additional sourcing: MIT Tech Review, FT, The Guardian, Ars Technica, The Register, Euractiv