Newsable: EU’s AI law looking to protect, without stifling innovation

Newsable is brought to you by BNZ.

It’s a question governments around the world are grappling with right now: how do you mitigate the potential harms of artificial intelligence, without hampering the potential benefits?

Well, the European Union has taken its first step into this vast unknown with a world-first ‘AI Act’ approved by European Parliament last month – a draft piece of legislation looking to shape global standards in AI regulation.

Nello Cristianini is a professor of artificial intelligence at the University of Bath, and tells Stuff’s daily podcast Newsable the legislation’s been a long time coming.

How would you describe the draft law?

“It was quite a difficult thing to do. It started quite a few years ago, and the fine line was: we want to be safe with artificial intelligence, but we don’t want to stifle innovation.

“This was a difficult line to draw, and different countries will do it in different ways. In Europe, the decision has been to not regulate AI per se, but to regulate its different uses. So the same algorithm could be allowed in a video game, but then in a bank it could be heavily regulated. That’s the fundamental idea.”

“It divides the different applications of artificial intelligence into levels of risk – it doesn’t go into the details of how you’re going to do it, so it allows for progress and change in technology. It just says, if you are going to make use of some AI to recognise the face of people in the streets, real time, that is going to be (an) unacceptable risk and not allowed in Europe.”

AI is often viewed through the prism of: this is an existential risk to humanity if it gets out of hand. You don’t have this view – you compare worrying about that to worrying about an asteroid hitting the planet, and argue that while that’s possible, a more pressing threat is pollution. Can you elaborate on that?

“That is my concern exactly. When I go to bed at night, I don’t stay up because of the extinction risk, but I do worry a lot about young people spending time in front of (the) recommender engine … getting their videos and their news from apps.

“I don’t know that anybody (has) studied properly what is the effect on a young mind of an algorithm like this. So that is a kind of risk that is not theoretical.

“There are billions of people – as an example there are 3 billion users, active users of YouTube – so many of them are young children, and they learn about the world through these tools. The tool is built to learn quickly what will make them click. I don’t want to point fingers at any company, but I think there is something to be studied here which is not future, it is present. We need to understand the effects on society of this. That is important.”

How big a deal is this, internationally?

“The first mover, I think, can set standards.

“Other countries will have very different interests – for example, in the United States, maybe corporations have a different set of priorities. Maybe in China, the state has a different relation with personal data.

“It’s a matter of establishing the values of the European mindset. And for us, it has been a decision in the parliament that keeping citizens safe, and keeping safe a set of principles – like equality, for example, and individual freedom – is as important as enterprise.

“But I do realise it is only one perspective and I’m not suggesting everybody should have the same values.”

Newsable is Stuff’s daily news podcast, wrapping up what’s worth talking about in a short package every weekday morning. You can find new episodes and more detail on our stories here or in our newsletter. Make sure to like and follow us wherever you get your podcasts and across Instagram and TikTok.