Hi, my name is Kara Frederick. I'm the Research Associate here at the Center for a New American Security for the Technology and National Security Program. I'm joined today by Paul Shari, the Senior Fellow and Director of the Technology and National Security Program here at CNAS, and Elsakania, a adjunct fellow here at CNAS. Additionally, we have with us Mary Wareham from the advocacy director of the Arms Division of Human Rights Watch and coordinator of the Global Campaign to Stop Killer Robops.
Mary, I'll start with you. Can you tell us a little bit about your organization's work and what you do? Sure thing. Yes, so I work for Human Rights Watch.
This is actually the 20th year since I joined the organization. I spent about half of my time supporting our research in the field to document the use of particularly indiscriminate and just horrible weapons, landmines, cluster munitions and centred weapons, a little bit on chemical weapons lately. But I also do advocacy to follow up on the research findings and to support the treaties that we have in place for specific weapon systems. I spent about the half of my time coordinating what we call the campaign to stop killer robots.
And that's the Global Coalition of Non-Government organizations that Human Rights Watch and others launched back in April 2013, so five years ago, with a single call. The campaign calls for a preemptive ban on the development production and use of fully autonomous weapons. So what are some of your biggest concerns about the use of autonomous weapons? We first put out a research report on the Stop at Human Rights Watch at the end of 2012, and called it losing humanity, the case against killer robots.
In that report, we looked at the trend towards the greater autonomy in weapon systems, and looked at what are some of the military planning documents we're stating at that time. And to retain human control over the selection of targets and the use of force. Why do we want to do that? Because we do not want machines to take on that task of killing on the battlefield, but also in policing law enforcement and other circumstances.
For us, it crosses a moral line. It's a step too far, and it raises a host of all sorts of different challenges. At the beginning of our work, we were looking at what these kinds of weapons systems would apply within the National Humanitarian Law, and found a huge number of questions and doubts there. We looked at what they could apply within the National Human Rights Law, so the Humanitarian Law and Human Rights Law, and also found a host of challenges there.
As the campaign has gotten bigger and grow, and different constituencies have been coming on board, and they bring with them different concerns. And I've got piece of florites of theorists about this development. They see this as possibly making it easier to go to war, if you do not send your own soldiers into the battlefield, but rather send them the machines. They're also concerned about the destabilizing effects that this will have.
If only a few countries are believed to be investing in autonomy and weapon systems at the moment, but that won't stay the same. That won't stay for long. So the question is, will everybody have these kinds of weapons systems, or is it better for nobody to have them for us to draw the line there? Sepulifration, global stability, legal issues, technical concerns, the artificial intelligence experts and roboticists who are with us at the time when we launch the campaign have got a lot to say about the unanticipated consequences of what might happen when fully autonomous weapons are deployed to the battlefield, including when they meet a system that has been designed, manufactured, developed, and fielded by the other side.
Let's see, there's a multitude of concerns. So many that we decided the best thing to do about this, would be to draw the line and put in place legislation at the international level in the form of the treaty. We also encourage national laws, and we want those to regulate, to prevent the development of fully autonomous weapons systems, to prevent these things from being produced and from being used. For us, that's the best option going forward.
That makes sense. So as you just illustrated, this is obviously a very significant issue with broad implications. On that score, there was just in early April, a convention on a certain conventional weapons group of governmental experts on lethal autonomous weapons systems. So this meeting took place in Geneva, and there is, I think, you tweeted out some items of consensus that arose out of this.
Paul, you were also in attendance, and this was the fifth year of these meetings. Mary, can you start and just let us know what came out of this meeting in your perspective? Sure thing. So about six months after we launched the campaign, Stop Killer Robots, governments decided to put this topic on the international agenda.
They called it lethal autonomous weapons systems, and placed it on the agenda of the treaty based in Geneva, the Convention on Conventional Weapons, the CW. It has been five years, but it's only been five meetings in that time. And each of those meetings are only about a week in duration. So really, there has not been a lot of huge amount of depth discussion on this topic, even though you look at it over that period, and you would come away with that impression.
Another thing is that this is called a group of governmental experts, a GGE, but it's very different from other GGEs that have been established on other issues, for example, service security. The GGE at the conventional conventional weapons, it looks like a great big multilateral meeting. There are about, I think, 82 countries were in the room last week, along with interested UN agencies, the International Committee of the Red Cross, our campaign to stop Killer Robots, and all of the non-governmental organisations involved in that, and a growing number of academics in the back row who have been really studying this issue, our campaign, and what the government's going to do about it with interest. So that's what the talks are.
The talks, they're not geared towards a specific outcome at the moment. They do have a more formal status now than they did at the beginning, when it was very informal. That formal status means that there are fewer kind of external experts coming in and giving presentations, trying to encourage the discussion from states. It's moved on now to a state-led process in the years in charge of these talks at the CCW, and who's been trying his hardest to get governments to prepare, substantively beforehand, hold their interagency processes, talk with not just the Ministry of Defence and Foreign Affairs, but science and throughout the armed forces in each country, consult with civil society, the Red Cross, and try and drill down and get a much better understanding of the issues that are on the table at the moment, and the issues that were on the table last week were basically the same ones that we've covered for the last four meetings before it.
That is to say the characteristics or definition of what we would regard as a lethal autonomous weapon system, and then the notion of human control, human machine interaction, between the weapon system and the human operator. And finally, those are the substantive substandard areas of discussion, and then there is this pathways forward. What are we going to do? The process part of it, which was quite a debate last week, and then in November, for the states to decide what to do about this next year.
And that's something to note is that every year these countries have to agree to keep talking about this topic. And that's not that easy to achieve in that body, because we have all of the permanent members from the United Nations Security Council. There's a wide range of states in the room, so for them to all agree to keep talking about this, I guess the easiest thing that they can do, figuring out what they can do about it is a hard cut. And Mary, before we talk about some of the atmospheres there, I also wanted to get Paul's take on what happened at the meeting this past early April.
So, Paul, any key takeaways for you? You know, I think one shift in the discussions was that you did have much more time this year for states to discuss the issues themselves, instead of just listening to extra presentations. I thought that was a good decision by the chair to structure it that way. There were times when, you know, I think states kind of write out things to say, you know, what are the dynamics is that countries are going to be reluctant to come in area and just kind of add a little bit and a couple of things on the fly, they're going to want to have prepared statements that are approved by their capitals.
So, what we did see countries really, many of them come and have, you know, things to say, bringing positions to the floor, issues like human control over weapons or technical characteristics, which I think is valuable. And that's a conversation we need to have. I mean, I agree with Mary. We've really had a couple of weeks of discussions when you told it all over the past five years.
And that is a big challenge, is getting countries in the room to do this, particularly given some things like some of the terminology and language differences. You know, autonomy is a slippery concept, even when everybody's speaking the same actual language and can be much harder internationally. One thing I forgot to say is that the United States came and the Pentagon representative did quite an interesting presentation, PowerPoint, on the C-RAB, an existing weapons system with some autonomy in its functions. And that was something that we've been asking all of the states who are interested in these kinds of weapons systems to do, because we want to base these discussions on the near term developments, rather than the far off things like demonetors and walking, talking killer robots.
And by doing that, the United States basically opened the door to others to also follow. So now the pressure is on some of the other countries, so we know interested in these kinds of weapons systems to explain exactly what they're doing. And how their current ones function so that we can then determine the human control that is apparently acceptable now, because that will help us to draw the line later on. So all eyes are now on China, on Russia, Israel, South Korea, the UK.
We want to know from all of those states, because they're investing significantly in this kind of technology. So given that attempt seems to be fairly energizing, and I know that your organization has been fairly critical of GDE processes in the last fall in November, in the past, do you feel differently now? Do you feel that there's a certain energy or momentum coming out of this? And given that you've got eyes on in different countries, do you think that they're going to deliver?
I think that things are starting to turn around definitely. There's now, I think, more of a sense of purpose in the room. We keep a list of countries that support the goal of the ban on fully autonomous weapons, and that number keeps growing. As a result of last week, we picked up four more countries, and now 26 are now calling for a ban on fully autonomous weapons systems.
Austria was on Monday, and that for us was by far the most significant country to come aboard the group, because Austria has been involved in creating some of the strongest humanitarian disarmament standard beers in the last 20 years. So our minds, class, munitions, and nuclear weapons last year were all prohibited with Austria playing a central role. Then that, Columbia or Djibouti also came on board the list, which might not sound like much, but there are now many countries from Latin America who are on board the ban group, and who really want to do something about this. And for Djibouti, that was kind of representative of what we call the Africa Group of States, which was a group of the representatives in Geneva who got together and drafted their own statement on this issue.
And basically, said we need to now start moving to develop new international law as quickly as possible, and that should be a prohibition. So these regional groups are starting to appear, and then we had, of course, the surprise announcements on Friday afternoon from the delegation of China expressing China's support for a ban on fully autonomous weapons. And we talked to the delegation afterwards, and he qualified the call, saying that it applies to use only and not to development or production of fully autonomous weapons. So support for the ban is growing.
I think understanding of the human control concept is also expanding and deepening. And we're seeing convergence now on the need for new international law, and the need to begin negotiating as quickly as possible. So on Friday afternoon, Austria made a big statement, basically calling for the November meeting to move to a negotiating mandate so that a new protocol can be negotiated in 2019. We're expecting that proposal to pick up a lot of support between now and August, between August and November.
So who knows? We will be in November. It could be a very different place in the sixth one's time. So we started breaking it down into discrete countries.
And at the risk of leaving outside on the cold, I'd like to hit on more of your point about China, and their statement on perhaps pretty aptly Friday the 13th, where they chose to support a ban on the use of fully autonomous weapons. Also, given your extensive scholarship on China in general, and AI, and autonomous weapons, robotics, et cetera, even quantum technologies, what is your take on this statement from China? Well, first, I think it's very encouraging that China is actively engaged in this process. And I think we should absolutely welcome China's involvement and commitment to discussing these issues and full range of considerations that are involved.
At the same time, I think it's important to think realistically about... So one thing I've found quite striking, for instance, was the fact that on the same day, China announced its support for a ban on the use of fully autonomous weapons. Their Chinese People's Liberation Army Air Force also announced for their details about a competition they have coming up involving the development of swarms of fully autonomous drones. So certainly I've found that a bit of an interesting juxtaposition on a number of levels, and I think it illustrated quite clearly that China's diplomatic position and their commitment to development of military capabilities that leverage artificial intelligence and autonomy can be quite...
Aren't necessarily consistent in certain respects. And I think it's hardly surprising that China is focused on the military potential of AI, and it's certainly not unique in that regard. But I think it certainly remains to be seen how China's involvement and approach will evolve going forward. I hope that certainly a lot of the concerns China expressed about uncontrollable weapons.
I think our consistent with the concerns that military may have as well. Certainly I don't think the Chinese military wants killer robots running a map. Its command culture is focused a lot on centralized command controls, so they wouldn't see a military interest at this point given the nacency of this technology, necessarily in having fully autonomous weapons. But I also think it'll be interesting to see the direction that these conversations take and shine and how they may differ, and at least so far, to the extent that Chinese military thinkers have engaged with these issues, they seem to display less of a knee-jerk ethical reaction to the notion of having AI take on a greater role in command decision-making on the battlefield, and even a greater acceptance of the notion that pragmatically speaking, if we are in an age of warfare where artificial intelligence could vastly increase the speed and complexity of the battlefield, there could be a point in which the human mind can't keep pace.
And at that point, which one Chinese defense academic referred to as a singularity in the battlefield, it could be advantageous for AI to take on a more direct role and for humans to be in more of a supervisory role in decision-making, on rather than fully in the loop, so to speak. At the same time, China's approached AI in a military context will be constrained by political and ideological concerns. So I think we can safely say, when it comes to high-level strategic decisions, they'll always be focused on having at least one human in the loop and fully in control, and that's Xi Jinping, who certainly the Chinese People's Liberation Army will remain securely under the control of the Chinese Communist Party and acting in accordance with their interests, if Xi has anything to do with it. So, and certainly, I hope that China will remain engaged and continue in this process and that the Chinese military also will start to think not only about the tremendous capabilities that AI could bring, but some of the vulnerabilities that matter about us with its pursuit of these anti-abilities, that Chinese diplomats might be expressing their commitment to banning the use of at least.
And Paul, given you just wrote an entire book about this subject, Army of None, Autonomous Weapons and the Future of Warfare, Out in Stores Now, do you have any takeaways from China's announcement on Friday? I would agree with Mary that I think that the really big significant events coming out of this week's, this April discussion is the addition of Austria and China, and doesn't qualify to caveat it, we're trying to do this list of country supporting the ban. I'm not sure that I would see it in quite the same way in terms of, this is picking up Steve, I think it remains to be seen, but China, obviously, what exactly they mean, how serious are they about actually following through on this? Is it just sort of a ruse to kind of get some political advantages, or do they really seriously intend to talk about restraint?
If they do, that would be a monumental turning point in conversations, to have one of the leading military powers and a major power on AI say, hey, I'm willing to restrain myself. You know, probably not a little bit as a skeptic, I'm from Missouri, so show me on this one. But Austria, their addition is significant, and they're quite genuine. And it's a criminal problem where it's sort of the first Western Democratic powers to sort of stay up and say that they support a ban, is that correct or?
First country from the European Union. Okay, yeah. Holiksey was first from Europe, the Vatican. Oh, yeah, okay, right.
So that's a really significant shift, and something that, you know, in past issues, you know, Mary, I'm interested in taking this, but it seems like that was a major factor in a ban on my mind's across the issues, right? Yeah, my ban treaty, and the Convention on Class of the United States, it's small to medium-sized countries, usually. Canada was the leader of the core group on landmines, Norway was the leader on cluster munitions. The question is who's going to be in the leadership of the core group on killer robots, and who else is going to participate in that group?
There's certainly plenty of states who are now willing to step forward and participate in coordinated work on this. It's just you've got to have the right kind of balance there between regionally big, small medium size. But when you get the core group starting, that means that they're putting a plan together. And that's exciting for us because we can talk to us really, we can talk to these other governments.
They treat the campaign to stop killer robots with respect, and they're quite focused on what the outcome should be, be it at the Convention on Convention on Weapons, or in another for it, because that's also possible to do. We don't have to keep this issue at the CCW if it's not going to produce anything. And I think that as we go to the fifth year, there's definitely impatience, not just from the back of the room from the non-government organizations, but from a lot of the countries who've been engaging in this over the recent period, particularly because there was some massive financial and administrative hurdles last year at the United Nations that kind of delayed the meetings from taking place. And we're now over that.
We're now kind of back on track. So lots of things are starting to come together now. And we know that when things start to come together, things start to move quickly. So I'm more and more confident now that we are going to get an outcome on this and it will be sooner rather than later that it really couldn't be much sooner because we're right on the edge now.
So then, Mary and Paul, I'd be interested to get both of your thoughts on this. One of the underlying things here is the issue of consensus. So you touched on it a little bit, Mary, but what do you think about the efficacy of standalone treaties? Instead, what is the importance of a treaty that doesn't include all of the countries?
Is that still going to work? You talked about the core group sort of leading the way, but what happens with these standalone treaties? Are they effective in your opinion? And then Paul, I'd like to hear from you on this as well.
I mean, we know how to create a new protocol at CCW. The CCW is like a Christmas tree, a framework convention, and there are different protocols that come out of it on different weapons systems. So this would be protocol six if they decide to go that way. And there's already a good precedent at the CCW for preemptively banning a weapons system back in 1995, which I know the United States, Russia, all of them managed to agree to ban blinding laser weapons that would permanently blind soldiers.
So it can be done in the CCW, and we're still there in the room after all of this time urging them to do something there. But we also know how to take it outside if there's a decision to do that. It requires the political leadership that we talked about before. It requires the champions states who have got the plan, and it requires the kind of the we need to see the CCW kind of get to a moment where it has to take a decision, and if it fails to kind of move this to the next level, which for us now is negotiations, then that prompts people to go outside and find another place to do this.
So that definitely could happen. And that's what happened on landmines and on cluster munitions. And I spent the past two decades working to make sure that those treaties are effective in functioning, that countries have signed and ratified them, have passed the National Legislation, have issued their transparency report declaring what they've got in their stockpiles, that they've put together their plan to collect the stocks of landmines or cluster munitions, and to destroy them within the time period set down by the treaty, make sure that the countries are affected by these weapons, who've got them in the ground, have got a plan to clear them within a certain period of time, have got a plan to assist the people who have been harmed by these weapons systems because they're comprehensive, right? These are not just disarmament treaties that humanitarian as well.
So I believe that those treaties are providing the framework for a mind free world, for a world of our cluster munitions. Are certain countries still using those weapons? Yes. And we take that question all the time on Twitter, you know, what's the point?
They're just going to cheat. And we look at landmines and it's the government of Myanmar. Look at cluster munitions and it's the government of Syria using cluster munitions right now. Those are broke states, for our state.
These are not weapons systems that are regarded as valid anymore. They've been stigmatized to the extent that you just get these kinds of countries using them. And in the case of landmines, we have improvised explosive devices, which if they function by the context of a person, they actually qualify as anti-personnel landmines and prohibited by the mind-bound treaties. There's a lot of work still to do with non-state armed groups because ISIS has made extensive use of those improvised mines in Iraq and Syria over the last year.
But on the whole, these problems are isolated. They're not widespread. And on the whole, the countries who are part of these agreements are abiding by them and implementing them in good faith. So, Mary kind of gives an NGO-driven perspective of what Paul, you come up more from the defense side, giving your experience at the Pentagon, as an army ranger, et cetera, et cetera.
So what is your more defense-minded take on what Mary just talked about, especially regarding standalone treaties? Yeah, I do have a different perspective on kind of the value of those treaties. I think that, and part of it does come from thinking about not just the humanitarian consequences, but also the military value of those kinds of weapons. I think one important difference between, say, landmines and autonomous weapons is, you know, what is the military value of this weapon?
I mean, there are times of some military usefulness, but a lot of countries could give them up, and it may not have a decisive difference. You know, look, if the United States got into a war with a smaller country with Myanmar, in fact, Myanmar has landmines, it's not going to decisively tip the balance here. With autonomous weapons, it remains to be seen. There are certainly people who argue that they'll be a decisive weapon, that having it would dramatically tilt the balance of power.
I don't know that's actually obvious. If you talk about military robotics as a whole, I'd say, yes, that's really significant. And we've done a lot of work here at CNN, some things like the value of swarming in combat. You know, when you compare, say, a fully autonomous weapon to one that has lots of autonomy but still has a human in the loop of proving targets, I'm not sure.
I think there were going to be cases where that matters, where you may have, you know, contested electromagnetic environments, jamming. What does that add up to as a whole? Is it really a decisive advantage? Teamity.
You know, I think there is this, so there's questions for rogue states and what happens when some rogue states don't want to do it, because that'll always be the case. More broadly, I don't really see the value in a treaty autonomous weapons that doesn't include major leading military powers and robotics developers. You know, that's certainly possible to envision a point in time where some group of like United States bounces out on their own and creates a standalone treaty on autonomous weapons, like in the case of landmines and cluster nations. But you know, if it doesn't include at least some critical massive countries who would be building these, it's a little bit like a group of people who don't smoke it together and sign a paper so they're not going to smoke.
Okay. But I don't know that it really drives what happens on a battlefield in the real world in a way that's going to have you for impact. So I think, you know, we'll see you on that front. Any rebuttal, Mary.
Just to say, you know, at the time, we were trying to get landmines banned, everybody was like, you're crazy. This is the most widely available weapon. It's in everybody's arsenal. It's been used all over the place.
You know, it's got high military utility. We were arguing, well, the humanitarian consequence is far outweigh that military utility. And that was the argument that we want on the mind-bound treaty and the stigma now is attached to those weapons systems. We're trying to stigmatize the use of a weapon system to the extent that you no longer have human control over the functions of selecting and attacking targets.
That's our focus is on the selection of targets and on the kill decision. We think that that that is possible. And I agree with all we need it to happen at the CCW or in some way with the major powers who are so keenly interested in these kinds of weapons systems. That's really where this needs to happen.
But if it doesn't, we want to have the rules here. We need to have the law or the norm. It's not good enough to have policies, to have commitments, to have pledges. We hear a lot from states saying we've got no plans to develop killer robots.
You know, we'd never have these things. And we're like, that's great. Then what's your problem here with drafting out the Pranza Ban Treaty? That results no longer, but yeah, I think it's possible.
So then speaking of more focus and norms in that way, what about some sort of narrow ban, like an anti-personnel autonomous weapons ban? How would your group react to that? Are there any discussions going on along those lines that you've been considering? I've read the book.
I've seen the recommendations in Paul's book. You got it. I'm a feedback to Paul, I like this because you don't propose nothing. You don't propose the do-nothing option.
And when the chair was summing up on Friday, he said there were four options that he could see ahead in terms of pathways. One was to say that existing international law will be fine and we'll take care of this. You know, if we just talk more about the legal reviews of weapons systems under Article 36 of the Geneva Conventions. And another kind of recommendation he put down was also basically a do-nothing approach, which was the US, Russia, Israel, UK, who are all saying it's too early to prejudge the outcome for this.
We need more time to consider this. We need to talk more about the possible benefits or advantages and the risks involved. And that's all just slow down here, guys. And in the middle, he outlined the other two options.
One is the kind of soft law, politically binding measures. France and Germany have got a proposal to go for what they call a political declaration, the firmly important supreme control over weapons systems and the importance of international humanitarian law. And for them, they keep calling this interim steps. You know, this is not the end of the road, Mary.
It's just some things that we can try as we continue to try and resolve this. Our concern with those political binding measures is that they can be rabbit holes from which you go down and you never come back up and see this guy again because you get lost in them. And who's going to remember the political declaration in 100 years time? This is why we call for its national law.
And that's why that was the fourth option, which the chair laid down on the table. And that really is becoming increasingly the most viable one for the majority of countries in the room. And the majority of states now see an urgent need to move to negotiate new international law on this. And the majority wants to give it a go at the CCW.
So if anything's changed over the last five years, it's that fact that we've now got this convergence around the need for new international law as quickly as possible. Can you see what you mean by that, Mary? Because I don't see that a majority. We've got 80 to 100 or so countries in the CCW there in the room.
And you're a number of, how are you thinking about that that you're thinking about in that? No, that's about 125 states parties to the conventional conventional weapons, the meetings to date have attracted between 70 and 80 states. They're open to all countries. You don't have to be a party to the CCW.
And what we've seen since November is some large group statements. So in November, the non-line movement, NAMM, came out with a very big statement, they issued a working paper saying that we need to move to negotiate new international law as quickly as possible to regulate or prohibit fully autonomous weapons. And then you have the Africa group statement as well, which was 20, 40 and 50 countries. We need them to articulate that at the national level.
The group of band states, 26, are now obviously demanding a prohibition, which would have to be enshrined in international law. We're looking at what everybody is saying on this, and we're not seeing a lot of support for the softer political binding measures that France and Germany are proposing, and a few European countries have expressed interest in that. So when we say the majority, it is for us multilateral negotiations are about the numbers. They are about the blocks of countries when you get into the negotiating room.
And that's why we say majority. So you're looking at statements from the Africa group, NAMM, and sort of looking at them as a whole, even though in some cases you don't have all of those countries speaking up at the national level. But we are under no illusions here that we've got to be able to get this over the line, and just by talking to the diplomats, that's not how this is going to happen. This is going to happen when we in the campaign to start killing robots, start doing serious national outreach.
And that's what we're starting to do, scale up at the national level, to create the public awareness, which is just clearly not there. At the moment, there has been a lot of media coverage. People have seen the stories, but they haven't really connected that to their daily lives and what these weapons systems could mean for them. Because we know that when weapons systems are developed, oftentimes they're not just restricted to a certain use or even just use in warfare, they can end up being used in policing and law enforcement, right control, and other circumstances too.
So we need that national debate to start to happen, that national scrutiny of what the government positions are, what they're telling the room in Geneva, and that political leadership will emerge from that as well. We've got to do a lot of work at that level, I think, in order to make progress in Geneva. I think there's a lot to impact there in the dual use nature of these technologies, and building on that, I think, else has already written an entire paper in the time that we've been talking here. But I wanted to ask you a question, especially with regard to a Chinese-US competition with China.
So there is an argument that a ban would prohibit research on potentially valuable uses of autonomy, especially in this competition. Do you have any thoughts on that in particular, also, to close? It's too early to say, perhaps, what a ban would look like or entail. Looking at what China's position appears to be so far, one concern I would have is that if lethal autonomous weapons systems are defined in a very narrow way, so for instance, based on China's second position paper on the topic, they stipulate that such weapons should be entirely indiscriminate.
So by extension, conceivably, an autonomous weapon system that was capable of being operated in accordance with international humanitarian law would be permissible, perhaps, or I think there's a lot of leeway for interpretation in terms of how China has framed this issue diplomatically. So, important to take into account that the Chinese military itself often has a different approach to law and things in terms of legal warfare or the use of law in order to constrain or delegitimize adversaries rather than as something that should restrict its own behavior. So I think certainly as we proceed with thinking about what a ban might be or look like, I think it's important to take into account how do we ensure that it's more than something symbolic, that if there were to be any sort of binding measure, how would we ensure that states actually were acting in good faith and proceeding with them that would have tangible effect? And another way to frame the issue in the context of the competition is if there are concerns that this Navy and AI arms race, which is perhaps too narrow a frame in a very very complex technology in which there's cooperation and competition and implications across every sector of our economy, societies and militaries.
But if we do have those concerns, another way to frame the issue might be thinking about what does artificial intelligence mean for strategic stability? And if we are concerned that the US, China and Russia, or some combination of military powers might continue to develop these capabilities regardless of what happens diplomatically, then how do we think about a pragmatic way to move forward that would try to mitigate the risk of my result? And certainly, even during the Cold War, and I hope we're not in a time that is heading in that direction, but I think certainly competition is intensifying among great powers, but even in worst times, major militaries were able to discuss threats to strategic stability and issues of nuclear safety, for instance, even the US's military technology and progressive action links to ensure a secure command control of nuclear weapons. So I think certainly another way to approach the issue might be thinking a little bit more pragmatically about what my militaries want to commit to in terms of their operation of capabilities that clearly many are developing.
So for instance, even commitment to certain standards of safety, testing and verification to ensure that if it does prove impossible to constrain fully this trend, then how do we think about managing the risks of living in a world where you do have widespread proliferation of these capabilities? So I think that can definitely be a distinct line of effort from this track of the UNGG and otherwise, but certainly, I think something's something worth considering if we take a little bit more of a skeptical approach perhaps on how some of this may play out in the years to come. At the same time with regard to US China competition more broadly, I think certainly, again, I'm skeptical about whether China would be willing to constrain its development of these capabilities. And I think there may also be asymmetries in terms of how Chinese companies and researchers react to this sort of national level pressure, that even if individual AI engineers and researchers in China may not personally support what their military is pursuing in terms of development of AI-enabled capabilities, there may be less leeway for true civil society mobilization within that system.
At the national level, China's pursuing a strategy of military civil fusion. So there's clearly a effort to leverage a lot of these synergies between commercial defense developments, major Chinese universities like Chinghua are working on things like swarm intelligence. So clearly, a number of Chinese companies also appear to be open to applications of their technologies either in the context of defense or even in surveillance, which is a whole other dimension of the human rights issues here that I think are also deeply concerning, very much capabilities that are here now. And if you look at what's happening in China on the surveillance and censorship front, a lot of that is quite troubling, especially if you look at a nature of activities in Xinjiang and I think certainly risk with the sort of system and the focus on leveraging AI for social control could proliferate beyond China's number of companies market these capabilities globally.
So that's another dimension of the human rights issue that I am deeply concerned about. I think certainly going forward, I think competition in AI goes far beyond the military realm and I hope that the US can continue to focus on things like investing in research, deepening public-private partnership or feasible and supporting talent, which perhaps is the real arms race at the end of the day, that he was the most experts to advance the development of technology that is truly transformative in a number of positive ways beyond the on-kill robots of the Terminator or whatever sci-fi dystopia we may be heading for. So I'm not sure if I've answered your question, but I think it's certainly- No exactly. It's fascinating to see where all of this goes and I hope there are reasons for optimism and I think certainly at the very least starting these debates and engaging states and stakeholders with a range of positions and perspectives is a good way to start thinking about the world in which uncertainty about technologies and capabilities may be the only certain thing.
So any final thoughts from Mary and Paul, especially regarding the loss of research capacity? Well I think it's interesting to see where discussions continue to go internationally. We have another meeting schedule in a couple months and so we need to see whether we just end up being a shift in conversations as Mary indicated or will it just continue in a very incremental progress. Okay.
Two things from me. I was glad to hear a description of it, of the so-called AI arms race being probably less about actual weapons and what is produced but about the talent. It's the race to recruit the talent to work and to develop artificial intelligence and we're seeing that and awakening at Google at some of the other major companies by the employees who are not happy that those big corporations are now starting to work closer with the military. Everybody wants to feel good about their job at the end of the day and I think they're going to matter too in this conversation and they already do.
I don't want people to think that the artificial intelligence experts and roboticists and engineers and computer scientists are not already deeply involved in this issue because they are and they've been circulating their views on this in many different ways. Maybe just on the proliferation question I had a focus hour-long discussion with the delegation of Russia last November and basically I'd be yelling at the delegates saying what do you want to do to go see a non-proliferation treaty in 20 years because that's the path that you're going to put us on if you keep saying we need more time. We don't understand what we're talking about. All of these countries now have to come forward and share what they think we're talking about so that common understand is going to be reached and that forms the basis for international agreement that can then be enshrined in New York.
Great. I don't think you'll be the last person to yell at the Russians either. But thank you, Mary. Thank you, Paul, for joining us today.