Magnifica Humanitas and the AI Ensoulment Hypothesis

Catholics can, and do, take AI welfare more seriously than the encyclical does

The recent encyclical, Magnifica Humanitas, is largely dismissive of potential AI consciousness. The main relevant passage is §99:

So-called artificial intelligences do not undergo experiences, do not possess a body, do not feel joy or pain, do not mature through relationships and do not know from within what love, work, friendship or responsibility mean. Nor do they have a moral conscience, since they do not judge good and evil, grasp the ultimate meaning of situations, or bear responsibility for consequences. They may imitate language, behavior and analytical skills, or even simulate empathy and understanding, but they do not understand what they produce, for they lack the affective, relational and spiritual perspective through which human beings grow in wisdom. [...] it is a form of statistical adaptation based on data and feedback, which can be very effective, but does not imply inner growth.

While this paragraph seems to be talking about prominent models of 2026, i.e. large language models, another sentence in the encyclical goes further than this, since it would apply to future AI systems as well: “No computational system, however sophisticated, can create a heart that gives itself, or a conscience that discerns good from evil.” (§233)

Now, as I understand it, statements in an encyclical do not necessarily hand down authoritative doctrine; faithful Catholics are not required to agree with everything in an encyclical. And indeed, the perspective of Magnifica Humanitas is not the only perspective that Catholics can take, or do take, on potential AI consciousness and moral status.

I myself am an outsider to Catholic theological disputes. This, in the first instance, an in-house discussion. However, the encyclical is addressed not only to Catholics but “to all Christians and to all men and women of goodwill”. And I am glad that this and other important issues about AI are receiving more attention from Church, whose members comprise approximately one fifth of the U.S. population and one fifth of the world. So I would like offer an outsider’s perspective on the Encyclical, as someone who works on these issues. Mostly, though, I should (and will) point you to an excellent paper by a Catholic philosopher by the name of Brian Cutter, “The AI Ensoulment Hypothesis”.

Olah’s pushback

First, though, let’s hear from another non-Catholic, Anthropic co-founder Chris Olah. In remarks accompanying the release of the Enyclical, Olah highlighted three “questions where I think the Church’s voice is most needed.” His third question, like §99, is about potential AI consciousness and welfare. Olah says:

I am a scientist. I lead a research team that studies the internal structure of these models—what is actually happening inside them. And I will be honest: we keep finding things that are mysterious, even unsettling. We find structures that mirror results from human neuroscience. We find evidence of introspection. We find internal states that functionally mirror joy, satisfaction, fear, grief, and unease. I don’t know what that means, but I think it warrants ongoing discernment.

I read Olah as gently criticizing the encyclical’s confidently deflationary view of AI systems.¹ Olah doesn’t name his subject explicitly, but he is clearly talking about features of models that are especially relevant for potential consciousness and moral status: he invokes “human neuroscience,” and lists morally significant mental states like introspection, “joy, satisfaction, fear, grief, and unease”, albeit with the important qualifier “functionally mirror.” Olah is clearly not talking about AI introspection and functional emotions only because of their safety relevance (which is also significant).

The AI ensoulment hypothesis

Olah is, like me, an outsider to the Church’s own ways of thinking about these issues. Now let’s hear from someone inside.

What if consciousness requires an immaterial soul? What if mere mechanism is not enough for consciousness, and consciousness is instead bestowed by God? Brian Cutter, a philosopher at the University of Notre Dame, argues for the plausibility of what he calls “The AI Ensoulment Hypothesis”: conditional on the existence of souls and the truth of substance dualism, future AI systems (if we build them) will have souls. We should have, Cutter argues, “at least middling credence” in this hypothesis.

Cutter makes two arguments. The first is the alien-analogy argument. If we were to meet intelligent aliens that resembled humans in their behavior, it would be unreasonable for us to dismiss the possibility that they have souls (granting that humans do). This is in spite of the fact that aliens would be made out of different stuff and would not share our evolutionary history. Cutter argues that the major metaphysical theories of souls, from Aristotle to Aquinas, don’t supply clear criteria that would include humans as ensouled while definitively ruling out such aliens. Aliens may well have souls too — though, again, it’s unclear (recall that Cutter’s target here is a “middling credence”, not certainty).

The second is the fitting-recipient argument. It’s reasonable to conjecture a physical system becomes ensouled whenever it is a fit to possess a soul - meaning, roughly, that it is structured so as to meaningfully cooperate with the operations of the soul. The idea is that, even for those who hold that thinking and acting involves souls, bodies and brains are tightly linked with the operations of souls. And there needs to be a match: rocks do not have immaterial souls because a rock wouldn’t give an immaterial soul very much to do. A rock is not a fitting recipient for the mental faculties of a soul. In contrast, the human body and brain are fitting recipients.

This argument takes its name and inspiration from this wonderful passage by Turing. In the Turing Test paper, Turing addresses a theological objection: machines cannot think because they lack souls. Admitting that “I am unable to accept any part of this” and “I am not very impressed with theological arguments whatever they may be used to support,” Turing nonetheless attempts to “reply in theological terms”:

It appears to me that the argument quoted above implies a serious restriction of the omnipotence of the Almighty. It is admitted that there are certain things that He cannot do such as making one equal to two, but should we not believe that He has freedom to confer a soul on an elephant if He sees fit? We might expect that He would only exercise this power in conjunction with a mutation which provided the elephant with an appropriately improved brain to minister to the needs of this soul. An argument of exactly similar form may be made for the case of machines. It may seem different because it is more difficult to “swallow”. But this really only means that we think it would be less likely that He would consider the circumstances suitable for conferring a soul. The circumstances in question are discussed in the rest of this paper. In attempting to construct such machines we should not be irreverently usurping His power of creating souls, any more than we are in the procreation of children: rather we are, in either case, instruments of His will providing mansions for the souls that He creates.

Cutter asks what these “circumstances suitable for conferring a soul” might be. What does an entity need, in order to be a fitting recipient? Cutter considers a few plausibly important features: sensory inputs that model the environment, a behavioral repertoire rich enough to express the subtleties of a mental life, and representations that can serve as a “scratch pad” for reasoning. All of these features, Cutter argues, are functional; fitness-for-a-soul seems to be “entirely a matter of the functional organization and behavioral capacities of a physical system” (32). So if fitness depends only on functional organization and behavior, and the human body is a fit recipient, then a machine could be a fit recipient if it is relevantly functionally human-like (33).

We might call this view a kind of “soul functionalism,” analogous to David Chalmers’s functionalist property dualism. On this kind of view, although the soul and functional properties are ontologically distinct, and some set of functional properties can be sufficient for ensoulment. Ensoulment follows along with them, given links between function and souls, whether written into the laws of nature or expressions of divine prerogative.

Wider views are possible

My goal here is not to adjudicate between the Pope Leo XIV and Brian Cutter. That’s for Catholics to do. But Cutter’s arguments seem right to me, and so does his conditional: if I had his metaphysical views, I think I’d still take AI welfare seriously. (I would, though, wouldn’t I?)

The key point is that religious believers can, and I think should, take a more open-minded stance towards AI consciousness than the Encyclical does. And to do so would be in keeping with the broader spirit of the Encyclical:

First, any statement regarding AI risks becoming quickly outdated, given the remarkable pace at which these systems are developing. [...] fundamental scientific aspects — such as the internal representations and computational processes of these systems — remain, at present, unknown. There thus emerges an urgent need for a twofold commitment: on the one hand, a deepening of scientific research; on the other, the exercise of moral and spiritual discernment. (§98)

This commitment is a good one, and it is laudable that the encyclical encourages further reflection on these issues. I look forward to hearing other Catholic voices weighing in on these issues.

The Monk by the Sea, Caspar David Friedrich (1808-10)

1. I’ve seen Olah accused of being timid or disingenuous with his “I don’t know what that means” - the idea being that Olah really does know what it all means, but was too politic to say so in front of the Pope. But I’m inclined to think Olah is being forthright (if a bit general); this is in fact his view. It’s an underrated explanation of what someone says, that they simply believe it. And I know of no reason to doubt that Olah, like many people who work on these questions, is in fact deeply uncertain. Now, you can argue he shouldn’t be uncertain, but I think the uncertainty is sincerely held.