Microsoft president Brad Smith: Ireland is “well-positioned for future of AI”

Ciarán Hancock (0:03)

Hello, and welcome to Inside Business with Ciarán Hancock. My guest this week is Brad Smith, Vice Chair and President of Tech giant Microsoft, one of the most senior executives in that global giant. He was in Dublin this week to speak at the government's International AI Summit in the RDS. And in my interview with him, you'll hear him talk about the need for regulation of AI, globally and nationally, Ireland's potential role in the development of AI, how AI could change our lives and our workplaces, and also Microsoft's plans and ambitions to develop AI. Here we go.

Brad, welcome to Inside Business and thank you for joining us. You're here for the International AI Summit in Dublin, which has been hosted by the government. And you have a slightly unusual message, I think, for a business leader, in that you want more regulation, particularly around AI. Tell us about that.

Brad Smith (0:56)

Well, I just think the state of AI technology and the state of our industry is at a critical inflection point. We're almost at a fork in the road, so to speak. The power of AI, the capability of AI has moved ahead faster than safety. We need safety to catch up. And that's really my call. We need also to take the kinds of steps that will ensure that the benefits of AI really accrue across the economy, not just say to the AI Model Lab companies on the west coast of the United States.

I do think especially when it comes to AI safety, we're going to need some laws. One of the points I've been making is that right now, there are more safety rules for a carton of milk than there is for the most powerful AI on the planet. That does not make common sense in my view. That doesn't mean we should want regulation to stifle innovation or become bureaucratically heavy handed.

But I think we need a bit more of the rule of law to keep people safe.

Ciarán Hancock (1:59)

Okay, we'll come on to that in a moment. But I heard you saying in your speech on the stage that you don't want AI worms in your fridge. And we've had all sorts of warnings about how AI could take over your air fryer, your oven, your microwave and your fridge and so forth. But we've also had really dire warnings that it could kill us all.

Do you buy into that?

Brad Smith (2:19)

Well, I always think we're better off taking things seriously, even if we might think the risk of something truly awful happening is low. I think the real question is, are we going to just listen to some of the AI labs offer warnings about what could go wrong? Or are we all going to work together to prevent these problems from happening?

I think we need to do more to prevent these problems, including cyber attacks or biological attacks or other things that could be truly awful as well as other harms. We can do a lot, in my view. I think we need to stop talking so much and start acting a lot faster and better together.

Ciarán Hancock (3:05)

So what does that regulation look like? Is it at a national level? Is this at a regional level, let's say in Europe or the United States, et cetera, or is it global?

Brad Smith (3:14)

I think we're going to see multiple layers, but we have to put the right rules in at the right places. I think when we look at something like an AI safety break, the ability to say, turn something off or slow it down, the way you can with, say-

Ciarán Hancock (3:30)

Kill switch.

Brad Smith (3:31)

I wouldn't use that phrase, but I would call it a safety break. I would analogize to the circuit breaker in your home to stop or prevent an electrical fire.

You put it as close as you can to the user. You put it in somebody's home. What you often find in those kinds of areas is there are certain industry standards, and then the application of those industry standards are embodied in law. Sometimes it's municipal law, sometimes it's national law, sometimes it's say EU wide or US wide.

The more you have an industry standard, the easier it is because you tend to find that everybody ends up pretty much having almost the same law regardless of how they do it. You definitely need at some level international collaboration. We don't think about the fact that every day, if we are picking somebody up at an airport or getting on a flight ourselves, that commercial plane, that aircraft that you fly on was tested, evaluated and ultimately certified by a government agency. And if you are flying an airbus plane, it was certified in Europe. If you're flying a Boeing plane, it was certified in the United States, but there's a reciprocity regime because of an international treaty that was enacted even before the end of World War II. Mostly I would say this, we do this for lots of products. I don't think we should have quite the level of, call it, allergic reaction to the possibility that it might just be sensible to develop something for AI.