Jon Stovell’s Notebook


Some thoughts on artificial intelligence in relation to theological anthropology

On Feb 3, 2017, Jonathan Merritt published “Is AI a Threat to Christianity?” in The Atlantic. Merritt suggests that “AI may be the greatest threat to Christian theology since Charles Darwin’s On the Origin of Species.”

Merritt’s argument is built on several, not entirely consistent presuppositions, and it is only their confluence that makes AI an alleged potential problem. These presuppositions are:

  1. The idea that a soul is a “component” of a human being.
  2. The idea that the defining trait of human beings is our intelligence.
  3. The idea that a collection of algorithms could become the same sort of being that humans are if we just make the algorithms sophisticated enough.
  4. The idea that “salvation” means the preservation of the human soul beyond death.

Presuppostion (1) is one possible interpretation of what a soul is in Christian theology, and even the dominant one during the medieval and early modern periods, but hardly the only one. In today’s theology (and more importantly, since we are thinking in terms of futurology here, tomorrow’s theology) that interpretation is considered dubious at best.

Presupposition (2) isn’t a Christian notion to begin with. It’s a modern Western notion, born out of and feeding back into our technologically driven cultural narrative.

Presupposition (3) presupposes (2) and adds the belief that the human mind is at its root a complex math equation calculating itself. Leaving aside the question of whether there’s any good reason to think that it is true (I don’t think there is), this idea not only makes intellect the defining human trait, but reduces the human being to the human mind.

Finally, presupposition (4) is rejected by many Christian theologians, as Merritt himself acknowledges. And as Merritt appears to recognize, if one understands salvation in terms of the redemptive, eschatological transformation of creation as a whole, AI does not pose a grave theological problem. If a self-aware AI ever became a reality, it would simply be another participating component in God’s overall creation that is to be redeemed.

It is worth noting that (1) and (2) are actually inconsistent unless and until one adds (3). If the soul (whatever that is) is a component of the human being, then it isn’t what defines a human being any more than a thumb or a vestigial appendix is, for the very simple reason that it’s a part, not the whole. Adding (3) implies that the human being is “really” just the human mind. That makes it possible to reconcile (1) and (2), but only at the cost of treating the human body as superfluous to the definition of the human being and ultimately something that could be set aside without any significant effect on the “real” human being.

Of course, the idea that body and spirit are distinguishable and even separable has a long (and fraught) history within Christian theology. If one treats spirit, soul, and (in a further step) mind as synonymous, then one can open the way for the idea that human mind = human being—although just opening the way is still not enough to justify taking the step. But suppose one does think there is reason to take that step. To reduce the human being to only the human mind in this way is, of course, a reduction. Every orthodox Christian theologian, even those medievals most fixated on the enjoyment of the Beatific Vision by disembodied Christians in the afterlife, still insisted that the physical resurrection of the body had to take place in order for the salvation of the complete human being to occur. Indeed, Christian theology has a word for the idea that the human body is ultimately superfluous to the human being: heresy.1

Rejecting any of (1) through (4) would make AI a non-threat to Christian theology. But in the case of presupposition (3), not only can we reject it, but we must reject it. Christian theology insists that human beings are not just human minds. It might be fun to imagine human minds being swapped between bodies, or “digitally copied,” or, reaching further back, reincarnated, but it isn’t a Christian idea. Since the idea that artificial intelligences could become beings commensurate with human beings in religiously significant ways is based on a concept of humanity that Christian theology would reject—and indeed, already explicitly has rejected—it doesn’t seem that any AI development could ever amount to a threat to its consistency the way Merritt supposes it would. Put simply, Christianity would first need to abandon utterly the concept of physical resurrection before the development of AI could even pose a theological conundrum, let alone a threat.