Edited for readability and completeness
Toni Sims is a researcher at NYU’s Center for Mind, Ethics, and Policy, where she studies the consciousness, welfare, and moral status of nonhuman minds.
KB Miller: Hello, Toni.
Toni Sims: Hi.
KB: Thank you for your time.
TS: Thank you for reaching out. I’m excited to talk about these issues!
KB: Sure. We want to hear more about the Center… Where you’ve been, where you’re going… What your aspirations are.
TS: I work for the Center for Mind, Ethics, and Policy at NYU. It’s directed by Jeff Sebo [author of The Moral Circle]. He’s a philosopher. One of the things that is special about our center is that we’re at a university and that’s really great because we have… a big network here of some of the best philosophers in the world, excellent neuroscientists, and we’re in the Environmental Studies Department, where there are people thinking about other kinds of nonhuman minds and nonhuman rights… It’s a very good place to be to be starting to ask these questions about a new kind of nonhuman mind.
KB: Right… I think that’s exactly right. And you have a long… tradition at NYU of cross-disciplinary research… the marriage of philosophy and science has never been more important than it is now.
TS: Yes. And I actually went to NYU as an undergrad, and I was a philosophy major. And then I did a master’s degree at Georgia State, where I was in the neurophilosophy program. So some of my coursework was in cognitive science. So that has prepared me for this.
KB: Great. And the philosophy, what was your track in philosophy?
TS: I was very interested in philosophy of mind, and then I worked for a while in the effective altruism movement. I was at Animal Charity Evaluators, where I was doing animal ethics… there’s philosophy of mind that’s relevant to animal ethics.
KB: That’s wonderful… Do you see some movement in animal ethics? I think there’s general movement in rights of nature globally, if not so much in the United States. In animal ethics in particular, do you think that the current issues around nonhuman rights for AI or AGI or ASI has had a salubrious effect on the march of animal rights?
TS: That’s a good question. I don’t know that I’ve seen that yet, but it’s still very early for AI welfare research, but I hope so. I think that it’s all co-beneficial if the effect is to get us to distance ourselves from human exceptionalism and to think more openly about what other types of minds could look like, then yes, I can’t imagine that it wouldn’t have that effect.
KB: [At AGI Ethics News] We’re certainly, hoping that that’s the direction that we’re headed in as well.
And of course it’s a challenge for the law, in particular, to make that step. And there appears to be from your colleagues at the More than Human Life (MOTH) group at NYU Law.
TS: Yes.
KB: There appears to be a lot of movement toward suggesting a law that gets away from the standard reach of moral philosophy.
TS: Yes… This is a project that we’re working on also. We’re working on a report about legal personhood for AI systems, and the gist of the report is, well… not necessarily [that] we should give AI systems personhood now, or we should never do it, but really, that personhood is a bundle of different rights and duties under the law that can come apart, and some of those rights and duties, we have argued and Jeff [Sebo] has written — he co-wrote the Philosopher’s Brief on Chimpanzee Personhood — some of those rights and protections should really be extended to animals and very likely should also be extended to advanced AI systems.
“A lot of our work is about looking at properties, like consciousness, sentience, agency, and looking at how we can assess those capacities in an evidence-based way. But we also acknowledge that we may never know with absolute certainty if AI systems can have those properties.”
There are some states that are already beginning to ban AI personhood [e.g. Utah, Idaho, and North Dakota], which seems very premature, and we would hope that, we hope that our forthcoming report will make the case that we should not be doing that yet, and we should be talking about it more first.
KB: Okay. It’s such a rich area of discussion. We had a good visit with David Gunkel a few months ago, and… we’re quite interested in watching the progress of relational ethics. And again, we’re getting signals from the nonhuman rights [researchers] — the More than Human Life (MOTH) team — that relational considerations, because of the systemic and ecological approach in rights of nature, make a lot of sense for the world that is approaching AGI. Do you feel [the same]? Do you feel a kinship with the relational ethics? Or more perhaps with the properties [approach].
TS: A lot of our work is about looking at properties, like consciousness, sentience, agency, and looking at how we can assess those capacities in an evidence-based way. But we also acknowledge that we may never know with absolute certainty if AI systems can have those properties. And so I think that’s where our relationships with AI systems can play a huge role, and we know that, just as a fact of the matter, the types of relationships that people have with different entities plays a huge role in their moral treatment.
So we know that, when people interact with something more as a tool or as a mechanism, then they’re not likely to attribute a mind to it and not likely to extend moral consideration to it. Whereas when people interact with something as a companion, in a more relational way, they are more likely to attribute a mind to it and more likely to extend moral consideration to it. And there’s a lot of psychological evidence about that. I actually wrote my MA thesis on this idea a long time ago, before I was even thinking about AI at all.
One thing that Jeff has actually written about… he has a paper with Eric Schwitzgebel, called the “Emotional Alignment Design Policy,” and one thing that they argued is that we should actually think about this when we create AI systems. And to the extent that it’s a design choice – whether we make an AI system that people will interact with in a more commodity way versus a more relational way, we should make it that way to reflect the properties that the system already has, as far as we know. Just to reduce confusion and to encourage people to treat [advanced AI systems] morally.
KB: So… is the underlying premise then that an LLM has a mind?
TS: Our underlying premise for most of our work is that we may never know with 100% certainty if it has a mind, but we want to have humility about it, and we want to be constantly reassessing and reevaluating and behave in a way that is proportional to the evidence.
The Center’s recent report Studying AI Welfare Empirically will be discussed in Part Two.
AGI Ethics News covers the ethical and societal implications of artificial general intelligence (AGI). We bring expert insights and thoughtful analysis to help readers understand the challenges and responsibilities of emerging AI technologies.