In the mid-twentieth century, Hannah Arendt wrote extensively about the critical importance of learning to think in the aftermath of a great rupture in our tradition of thought. She wrote of the desperate situation when “it began to dawn upon modern man that he had come to live in a world in which his mind and his tradition of thought were not even capable of asking adequate, meaningful questions, let alone of giving answers to its own perplexities.”
Frequently, Arendt linked this rupture in the tradition, this loss of a framework that made our thinking meaningful to the appearance of totalitarianism in the early twentieth century. But she also recognized that the tradition had by then been unravelling for some time, and technology played a not insignificant role in this unraveling and the final rupture. In “Tradition and the Modern Age,” for example, she argues that the “priority of reason over doing, of the mind’s prescribing its rules to the actions of men” had been lost as a consequence of “the transformation of the world by the Industrial Revolution–a transformation the success of which seemed to prove that man’s doings and fabrications prescribe their rules to reason.”
Moreover, in the Prologue to The Human Condition, after reflecting on Sputnik, computer automation, and the pursuit of what we would today call bio-engineering, Arendt worried that our Thinking would prove inadequate to our technologically-enhanced Doing. “If it should turn out to be true,” she added, “that knowledge (in the modern sense of know-how) and thought have parted company for good, then we would indeed become the helpless slaves, not so much of our machines as of our know-how, thoughtless creatures at the mercy of every gadget which is technically possible, no matter how murderous it is.”
That seems as good an entry as any into a discussion of Lethal Autonomous Robots. A short Wired piece on the subject has been making the rounds the past day or two with the rather straightforward title, “We Can Now Build Autonomous Killing Machines. And That’s a Very, Very Bad Idea.” The story takes as its point of departure the recent pledge on the part of a robotics company, Clearpath Robotics, never to build “killer robots.”
Clearpath’s Chief Technology Officer, Ryan Gariepy, explained the decision: “The potential for lethal autonomous weapons systems to be rolled off the assembly line is here right now, but the potential for lethal autonomous weapons systems to be deployed in an ethical way or to be designed in an ethical way is not, and is nowhere near ready.”
Not everyone shares Gariepy’s trepidation. Writing for the blog of the National Defense Industrial Association, Sarah Sicard discussed the matter with Ronald Arkin, a dean at Georgia Tech’s School of Interactive Computing. “Unless regulated by international treaties,”Arkin believes, “lethal autonomy is inevitable.”
It’s worth pausing for a moment to explore the nature of this claim. It’s a Borg Complex claim, of course, although masked slightly by the conditional construction, but that doesn’t necessarily make it wrong. Indeed, claims of inevitability are especially plausible in the context of military technology, and it’s not hard to imagine why. Even if one nation entertained ethical reservations about a certain technology, they could never assure themselves that other nations would share their qualms. Better then to set their reservations aside than to be outpaced on the battlefield with disastrous consequences. The force of the logic is compelling. In such a case, however, the inevitability, such as it is, does not reside in the technology per se; it resides in human nature. But even to put it that way threatens to obscure the fact that choices are being and they could be made otherwise. The example set by Clearpath Robotics, a conscious decision to forego research and development on principle only reinforces this conclusion.
But Arkin doesn’t just believe the advent of Lethal Autonomous Robots to be inevitable; he seems to think that it will be a positive good. Arkin believes that the human beings are the “weak link” in the “kill chain.” The question for roboticists is this: “Can we find out ways that can make them outperform human warfighters with respect to ethical performance?” Arkin appears to be fairly certain that the answer will be a rather uncomplicated “yes.”
For a more complicated look at the issue, consider the report (PDF) on Lethal Autonomous Weapons presented to the UN’s Human Rights Council by special rapporteur, Christof Heyns. The report was published in 2013 and first brought to my attention by a post on Nick Carr’s blog. The report explores a variety of arguments for and against the development and deployment of autonomous weapons systems and concludes, “There is clearly a strong case for approaching the possible introduction of LARs with great caution.” It continues:
“If used, they could have far-reaching effects on societal values, including fundamentally on the protection and the value of life and on international stability and security. While it is not clear at present how LARs could be capable of satisfying IHL and IHRL requirements in many respects, it is foreseeable that they could comply under certain circumstances, especially if used alongside human soldiers. Even so, there is widespread concern that allowing LARs to kill people may denigrate the value of life itself.”
Among the more salient observations made by the report there is this note of concern about unintended consequences:
“Due to the low or lowered human costs of armed conflict to States with LARs in their arsenals, the national public may over time become increasingly disengaged and leave the decision to use force as a largely financial or diplomatic question for the State, leading to the ‘normalization’ of armed conflict. LARs may thus lower the threshold for States for going to war or otherwise using lethal force, resulting in armed conflict no longer being a measure of last resort.”
As with the concern about the denigration of the value of life itself, this worry about the normalization of armed conflict is difficult to empirically verify (although US drone operations in Afghanistan, Pakistan, and the Arabian peninsula are certainly far from irrelevant to the discussion). Consequently, such considerations tend to carry little weight when the terms of the debate are already compromised by technocratic assumptions regarding what counts as compelling reasons, proofs, or evidence.
Such assumptions appear to be all that we have left to go on in light of the rupture Arendt described of the tradition of thought. Or, to put that a bit more precisely, it may not be all that we have left, but it is what we have gotten. We have precious little to fall back on when we begin to think about what we are doing when what we are doing involves, for instance, the fabrication of Lethal Autonomous Robots. There are no customs of thought and action, no traditions of justice, no culturally embodied wisdom to guide us, at least not in any straightforward and directly applicable fashion. We are thinking without a bannister, as Arendt put it elsewhere, if we are thinking at all.
Perhaps it is because I have been reading a good bit of Arendt lately, but I’m increasingly struck by situations we encounter, both ordinary and extraordinary, in which our default problem-solving, cost/benefit analysis mode of thinking fails us. In such situations, we must finally decide what is undecidable and take action, action for which we can be held responsible, action for which we can only hope for forgiveness, action made meaningful by our thinking.
Arendt distinguished this mode of thinking, that which seeks meaning and is a ground for action, from that which seeks to know with certainty what is true. This helps explain, I believe, what she meant in the passage I cited above when she feared that we would become “thoughtless” and slaves to our “know-how.” We are in such cases calculating and measuring, but not thinking, or willing, or judging. Consequently, under such circumstances, we are also perpetually deferring responsibility.
Considered in this light, Lethal Autonomous Weapons threaten to become a symbol of our age; not in their clinical lethality, but in their evacuation of human responsibility from one of the most profound and terrible of actions, the taking of a human life. They will be an apt symbol for an age in which we will grow increasingly accustomed to holding algorithms responsible for all manner of failures, mistakes, and accidents, both trivial and tragic. Except, of course, that algorithms cannot be held accountable and they cannot be forgiven.
We cannot know, neither exhaustively nor with any degree of certainty, what the introduction of Lethal Autonomous Weapons will mean for human society, at least not by the standards of techno-scientific thinking. In the absence of such certainty, because we do not seem to know how to think or judge otherwise, they will likely be adopted and eventually deployed as a matter of seemingly banal necessity.
Update: Dale Carrico has posted some helpful comments, particularly on Arendt.