Paul Murphy: Military AI must never decide life or death

for the presentations. Firstly for Professor O'Sullivan, just a question which is not particularly AI related but it arises from your opening statement. You speak, in December last we saw a Russian dark ship using drones to fly into the flight path after Zelensky had been there. What evidence is that there was a Russian dark ship? This is the view of the defence forces, that this was a Russian dark ship. So it's known that there were Russian dark ships operating in the Irish Sea at that time. I'm not aware of any statements by the defence forces, you could be right, but saying that it was a Russian dark ship. You're making it public now. That goes much further than any minister. Initially ministers said that the use of the drones was suggestive of Russian involvement, because obviously one can guess who might have an interest in it, but then actually they retreated from that more recently and didn't repeat it. But you're saying defence forces have the information that it was launched by a Russian dark ship? Yes, to the best of my knowledge that's the case. Okay, that'll be very interesting, we should explore that probably not in this committee, but with the defence forces. Just to go to the AI, and obviously you're a member of the Global Commission on Responsible AI in the military domain, some might argue that that's a contradiction in terms. You might come back to my first question and incorporate that. I have a question, the most high profile recent use of AI in military conflict has been Israel's use in the genocide of Palestinians in Gaza. A whole number of AI-fuelled operations called the Lavender, the Gospel, Where's Daddy, where effectively they used AI to identify, anyone identified as an even low-ranking mass operative, came up with something like 38,000 people. According to reporting by the Plus 972 magazine, they checked for accuracy, they found that 90% of these are accurate, and then on that basis went ahead with a fully automated system. So even, I mean, someone being a mass operative doesn't justify their assassination, and then on top of that, you had, they authorised, they said we will greenlight these operations, AI will greenlight these operations, even if for each low-level mass operative we kill 10 to 15 civilians. Do you have a view on this use? Does this account, does this amount to responsible use of AI? No, it doesn't. So I guess, so first of all, I should make absolutely clear, I'm absolutely no fan whatsoever of the use of AI in the military space. In fact, I've spent the last ten years trying to hold back the use of AI in those spaces. And the reason why you have commissions like GC Re-Aim and others is that essentially civilian legislation like the EU AI Act is effectively silent on these things. And there does, we do need to have some basis for discussing, describing, governing, and ideally legislating in that space. But unfortunately the most powerful actors we have internationally who have capability in this area, and you mentioned Israel, they certainly are very strong. They have a very high level of technological capability. They don't tend to really participate as positively as you might wish with the UN processes. So when we talk about these Track 2 issues, the GC Re-Aim, it's not that one is saying that there are, that AI being used in the military space is responsible. We're trying to find ways in which one can ultimately get to a treaty or a piece of legislation globally. AI systems should never take decisions about the life of individuals. The, when you hear, and this is something that often comes up when people describe the capability of AI systems, well, you know, my AI system is 90% capable or 99% capable. When it comes to the risk to human life, that's not adequate at all, right? So, and ultimately the responsibility for taking decisions over people needs to be taken with great responsibility. And of course, what does that mean? And what you'll find when you speak to, when you speak to in Track 2 world, you don't really speak to formal representatives of governments, but you speak to people who have been in government possibly or in the defence forces or whatever the case may be. And when you talk to them about their thinking on it, it becomes clear what these principles are. But of course, what's happened in the Middle East is absolutely abhorrent, you know, and the use of any sort of technology against the civilian population is just, it's beyond, it should be beyond, you know, human consideration, absolutely. So I mean, related to that, you said there's broad international support for the banning of unpredictable AI systems, e.g. online learning systems, fully autonomous human targeting weapon systems and fully autonomous nuclear weapon systems. I mean, kind of one would hope so. One would hope so, yeah. Are there states that are opposed to or certainly haven't endorsed the idea of banning fully autonomous human targeting or fully autonomous nuclear weapons? Yeah, so for example, so when you, so again, if you look at what's happening with the GGE in laws, these, you don't tend to find great agreement around very much here, but when you talk to individual representatives of militaries internationally, I've yet to meet a military person who has anything, who says anything other than these things should be absolutely banned. But I presume kind of related, I presume Israel just doesn't say anything about it, a bit like they don't support nuclear non-proliferation treaties. What's missing is, what's missing is the mechanism and the structure for forcing that debate, and unfortunately that isn't likely to be the UN, because you know, China, Russia, United States and indeed Israel, you know, they have a history of not... And what is the position of all the big powers, let's say US, Russia, China, UK? So there are track two dialogues that are going on, so there are these informal agreements in place about the use of the, about the use of these technologies. And of course, you know, it's not completely a Wild West space, there is international humanitarian law and so on that does govern it. But I think in my view, the world is calling out for some structure, some organization that can step in and have that debate. And I'm not advocating that this be Ireland simply because I'm sitting here in the Oireachtas or that I'm Irish, but I think we really do have a very special position internationally, and I think we have that trusted position, and I think it would be a great contribution to international peacekeeping and so on, that we did drive that. And you know, there are, you know, so GC re-aim, while it's a group of experts who've come together and sort of spec'd out, well, what does a responsible version of AI look like in the military space? I should add that, for example, the United States hasn't signed up to what the GC re-aim has advocated. So there's a lot of work to be done, but I think in the discrete track two world, there is opportunity, and I think Ireland could really step into this, could really step in and lead there.