ОБ ЭТОМ ЭПИЗОДЕ
Contesting Paul Scharre’s influential vision of “centaur warfighting” and the idea that autonomous weapon systems will replace human warfighters, Sparrow and Henschke propose that the manned-unmanned teams of the future are more likely to be minotaurs—teams of humans under the control, supervision, or command of artificial intelligence. They examine the likely composition of the future force and prompt a necessary conversation about the ethical issues raised by minotaur warfighting.
Keywords: ethics, manned-unmanned teaming, future force, centaur warfighting, autonomous weapon systems
Read the article: https://press.armywarcollege.edu/parameters/vol53/iss1/14/
Episode Transcript: “Minotaurs, Not Centaurs: The Future of Manned-Unmanned Teaming”
Stephanie Crider (Host)
You’re listening to Decisive Point. The views and opinions expressed in this podcast are those of the authors and are not necessarily those of the Department of the Army, the US Army War College, or any other agency of the US government.
Today I’m chatting with Rob Sparrow, professor in the philosophy program and an associate investigator in the Australian Research Council Center of Excellence for Automated Decision-making and Society at Monash University, and Adam Henschke, an assistant professor in the philosophy section at the University of Twente, Netherlands.
Sparrow and Henske are the authors of “Minotaurs, Not Centaurs: The Future of Manned-Unmanned Teaming,” which ran in the Spring 2023 issue of Parameters. Welcome to Decisive Point, gentlemen. Set the stage for us, please, including Paul Sharre’s perspective.
Robert J. Sparrow
We’ve seen drones and teleoperated weapon systems play an increasing role in contemporary conflict. There is lots of enthusiasm for autonomy in weapon systems in the military, as I think 20 years ago, there was a paper in Parameters arguing that in the future the tempo of battle would increase to such a point that only computers would be capable of making the decisions that are required to win battles. So for a long time, there’s been a debate about the relationship between human beings and unmanned systems in warfighting. Recently, Paul Sharre has argued that we don’t need to worry about autonomous weapon systems taking over all the combat roles because, actually, the future of warfighting involves manned-unmanned teaming, and Sharre suggests that we should think of this on the model of what he described as a Centaur. A centaur is a mythical creature with the head and upper body of a human being and the lower body of a horse. And that’s a really nice image. You’ve got the human being in command and in control and the machines doing the physical work involved in warfighting.
We think that’s perhaps optimistic for a number of reasons, and that what we’ve seen, is really in applications, is it’s often easier to get machines to be making decisions than it is to get machines to do physical work. And so, for that reason, we think that the future of manned-unmanned teaming might be better imagined as what we call a minotaur. And so, rather than the human being in charge of the team, we suspect that in many roles, actually, the AI will be in charge of the team. And the human beings will be effectively under the command of the AI and doing the physical work, where the mental work will be performed by artificial intelligence.
Henschke
One way to think of Paul Sharre’s approach is, as Rob said, he’s advocated this view of centaur warfighting, and there the human is generally seen as the head, the decision-making part of the warfighting operation. And the robots and the machines, they do the, kind of, the grunt work. They’re the things that do the stuff on the ground. So, the way in which this manned-unmanned kind of vision is put forward in Sharre’s work is humans do the deciding and the robots, or the machines, do the fighting
Host
But you disagree. Will you expand on that please?
Sparrow
We do think that in many domains and in many situations it’s more likely that the machine will be in charge—or effectively in charge. We think that it’s quite hard to get machines to do something like move a gun into place or walk up the stairs or talk to these people and ask them where the insurgents are. Those roles, we think, will still need to be carried out by human beings. But, for instance, wargaming or identifying and tracking targets, those tasks machines can outperform human beings already in lots of circumstances.
If you are considering the team, we think a lot of the executive and cognitive tasks will actually be handed over to artificial intelligence, and the human beings will be left doing what the AI says. And so, in that context, you should think of this as a minotaur team, as a kind of cyber with an artificial intelligence head and the body made-up of human beings.
Henschke
On this, Rob came up with the idea of flipping the central view to suggest that we might instead think of these robot/human teams like minotaurs, where the thinking is done by the machine and the fighting, or the grunt work, is done by the humans. So, the robot becomes the head, and the human becomes the body, kind of flipping it from the central vision. The idea is that we ought to think of these more as unmanned-manned teams, like a minotaur rather than a centaur.
Host
I love that visual. Would you walk us through the key technical dynamics and the particulars of minotaur war fighting?
Sparrow
When people started thinking about artificial intelligence, they thought the hard task was going to be getting machines to do the things that we find difficult—playing chess, calculating, looking for patterns in data. But when people started building robots, what they discovered was the machines were quite good at that stuff. Machines were able to calculate . . . were able to do scheduling tasks. They were able to, in the end, play chess reasonably easily and very, very well.
Where machines struggled were doing things that children and animals can do, and so we don’t recognize as requiring sophisticated capacities— things like walking into a room and recognizing where the chairs in the room are or simply being able to walk up steps or pick up a cup. Those tasks actually turn out to be incredibly difficult for machines.
We still don’t have robots that can, for instance, walk into your office and find your coffee cup. Machine vision systems are much better but in complex environments, and where there’s a need to recognize context and maybe move around the cluttered environment, machines fail very quickly. If you’re looking for tasks to automate, if you’re looking for tasks that machines can perform, often where people end up employing machines is in these executive or cognitive roles. For instance, scheduling which offices get cleaned and making sure that there are people to clean them . . . that task can be outsourced to a machine scheduling system. Actually walking into the office and emptying rubbish bins, setting for the next vacuuming, that still needs to be done by human beings.
Host
Let’s talk about ethics. What are the ethical implications of minotaurs and minotaur warfighting?
Henschke
Well, one of the main things in the notion of the minotaur warfighting as we describe it in the paper is that you’ll have the machines, computers, and AI in a combination of technologies, directing and guiding how humans engage in a conflict zone. And as result of that, you’ve got decisions being made by machines carried out by humans. And this has quite a few ethical implications. One of the most interesting ones, or at least one of the first ones that we want to point out, is in these situations there might actually be a strong case for minotaur warfighting. If this would either decrease the likelihood of military mistakes, would decrease the likelihood of fratricide, and/or put your own soldiers at unnecessary risk, we might want those decisions to be made by the computers, by the machines. And so, there is a bit of an argument in favor, first, of minotaur warfighting, and, picking up from what Rob had just said, there might also be a case for minotaur warfighting in many situations over centaur warfighting.
So, if you think of what Rob was saying about navigating physical terrain, when we think of terrestrial warfighting—warfighting on land, for instance—that’s really really complex, really complicated physical environments. You might have sand. You might have water. You might have trees. You might have humans or equipment moving through these complicated environments. And in that sense, the centaur is probably gonna face a whole lot of trouble. Whereas the minotaur system, or minotaur unmanned-manned system, that might actually have a far better capacity to operate in these sorts of complicated, complex, terrestrial environments. So, there we might see that there’s actually a case for the minotaur warfighting if it’s going to increase both the likelihood of success and decrease the chances of making military mistakes.
Sparrow
We also think that people might be quite horrified by that prospect. I mean, really, the idea that you were just following a list of tasks or strategic objectives given to you by a computer system, I think people are going to really struggle with that. And there’s an understandable sense that human beings are valuable in a way that machines are not and that placing machines in authority or giving machines effective power over human beings is getting that relationship backwards.
We think there’s a relationship here with the debates about autonomous weapons systems, where critics of autonomous weapons systems have oft...
