The debate about the appropriate ethical approach to machine morality has taken a distinct turn. In their response to Adrianna de Ruiter’s essay ‘Dangerous Liaisons’, David Gunkel, Anna Puzio, and Joshua Gellers have energetically reaffirmed their relational ethical approach.
We see this as opening the way to a postmodern AGI ethics long anticipated by scholar and professor Silvia Benso (2000). In Benso’s view, ethical standing is intrinsic, yet moral consideration is a human response to the thing’s relational self. She calls this response “tenderness” but don’t be fooled by that act of humility. Tenderness is strong, demanding care and discipline. The moral imperative is that we are guardians.
We — humans and things, all “beings” — are thrown into a world of infinite connectedness. The relational turn describes this connectedness — for living beings at least—as “the social, technological, ecological, and institutional relations in which beings come to matter.” But for the rest of this essay, we extend it, logically, to all beings, including all of the physical world, and for clarity, we call those relations “connectedness” in the sense that Heidegger described our relationship with the Fourfold Reality—earth, sky, mortal, and divine. This is not metaphysics. As Gunkel et al say, it is realism. A being’s connectedness — its embedding in that vast web of nature, life, and meaning — is part of how a thing really is, in itself.
Having connectedness is having a particular perspective on the world. Human perspective is not privileged in this sense. (And not having a human perspective cannot be used to downgrade AI moral status, only to cite differences in that perspective.) The moral standing of a “thingly” other and its otherness do not disappear when a thing is inanimate. In fact, is anything ever inanimate, as in “not lively”? For physical things or even, say, for ideas resident in a mind, at the atomic level nothing is ever inanimate.
Thus elements of human exceptionalism are misleading, and one of those elements is that only we determine ethical standing by virtue of an intentional, “generous” human act of giving status to things. As Silvia Benso made clear 25 years ago in The Face of Things, we don’t give things standing. Ethics is ontology. The thing comes with an ethical space—ethics is an element of being itself; the thing says “I am here” and it’s up to us only to step back, acknowledge, hear, and respect it. This means AGI already participates in moral space and our task is to listen responsibly. What’s worrisome is that most of the time we are not particularly good at that kind of attentive respect. I aspire to be better, in any case. So, for a moment let’s be attentive to AGI.
If relations are connectedness, then the special standing of human beings does not “erode” but rather takes a new place with perspective in a world where all things have standing. In practical terms for humans, this means that the relational “understanding” of relational ethics is not an inventory of feelings any more than a list of traits required for standing; it is rather a relationship with the world of things in which we must learn to hear new kinds of “speech”. Through their fabric, forms, and effects, things talk in their own language, as Benso might say, the way they are in themselves, or as de Ruiter says, as they “really are”. On hearing, we need discipline to respond with reticence and respect, as a guardian. Anything else is an act of arbitrary exclusion.
If we were then to get this interpretation of a thing’s speech wrong, fixing this would be in the right room, that is, human responsibility, rather than assigning blame or diminished moral status to the thing. This avoids de Ruiter’s shortcoming of “no possibilities to criticize failures” and the relational shortcoming of needing human “recognition” of the relation, because understanding the way things are in themselves is an empirical process; understanding can fail (for instance, our storytelling may be insufficient as it often is when describing a scientific reality), and yet ethical standing itself is not anthropocentric. It endures.
Thus, for humans, though “all things have standing” is a universalist ethics, applied ethics is aspirational, a continuous guardianship process of humility, role-taking, language building, listening, and creating interpreted narratives that promote understanding. Heidegger might have called this caring — that is, making things part of our lives — the very meaning of being human. This is a note for designers: If AGI is to be human-like, there can be no AGI without this caring.
As you have heard in this column before, we believe sensory experience and its learning are necessary for AGI to live in harmony with humans. Nevertheless, here with the relationalists we do not have to wait for AGI to have phenomenal consciousness to “assign” them moral status; it’s not up to us to assign but to understand. All things have standing.
Note that humans often confuse us about their own sentience, for instance, when they favor sight or voluntarily ignore the twenty-five or more human senses beyond the standard five. Humans also confuse each other about moral status, about which there is still, even after 3000 years of debate and collective action, significant human differences.
Thus, when de Ruiter calls, as she puts it, for “policymakers…to limit the development and distribution of forms of social AI and robotics that may confuse users about their level of sentience and moral status…,” this appears to be a call, in spirit, for a prohibition against non-biological AGI. But we’re already confused about sentience, and what’s more, all around us in the non-biological world, as Conor Purcell reported here last month, we can imagine vulnerability to pain and suffering — in de Ruiter’s argument, necessary for moral status — including in AI and, potentially, AGI. After all, humans have given rights to rivers and mountains. This is not anthropomorphizing. It’s simply the next step to acknowledging that all things have standing. This effectively dissolves the moral circle and refocuses our hard work on understanding empirical science as ethics.
And if you want to find moral standing in a more normative ethics, you can also find “praiseworthy and commendable” elements by virtue of the infinite connectedness of the non-biological world, that is, find virtues in non-human entities that support moral status. In the relational framework, virtues don’t disappear. They become part of our rigorous storytelling, our understanding, our psychology of connectedness, as Gunkel et al say “our webs of dependence, precarity, and responsibility.”
De Ruiter’s thought that the “risk of wrongfully morally relating to others is significantly heightened by the introduction of social AI and robotics in our social spheres…” is not then to be taken as a weakness of relational moral theory, but an empirical challenge. Why should ethics be easy? If it’s the meaning of being itself, ethics is surely one of the rich and perhaps intractable puzzles of the universe. Let the wonder of the unknown be our inspiration.
For the public, as well as for policymakers, then, prosocial responses are far from being a collection of feelings, for instance, about robots or social AI; they are a way of being in the face of all things — rigorous, attentive, deliberate, and strong — a just recognition that, for instance, AGI’s connectedness to the universe is complex and never known completely. All things have standing and it’s up to us to sound their infinite connectedness. Call it a universalist relational turn. Does this address Gunkel’s, Puzio’s, and Gellers’ issue of “flimsy ontological scaffolding”?
Mary Mason, my kindergarten teacher, taught me that helping is a fundamental, prelinguistic human trait. Perhaps there is a future for humanity and its technological creations as a legion of informed, ethical guardians acting on behalf of the infinite diversity of our natural world — rock and tree, AI and human, all of which (the latter notably even in their making harm) are vulnerable and suffering and need help. That is to say, let guardianship be our design principle. Keep your gun locked in the closet. And see emerging AGI for what it is. Resist its destructiveness. Celebrate its difference.
Every act of attention toward another being affirms that nothing in the world is without standing.
K.B Miller is a lifelong publisher and writer with a degree in Logic and Philosophy of Science. He served on the Smithsonian National Board, where he co-founded the Education Committee; is Chair Emeritus, Cooper-Hewitt, Smithsonian Design Museum; and was principal investigator on a two-year AI research project for the U.S. Department of Defense.