Skeptical theism is a cluster of responses to the problem of evil that appeals to human cognitive limitations. Consider this classic argument against the existence of God:

1. 1. If an all-good, all-knowing and all-powerful God exists, then this God is justified in permitting any particular evil only if that evil is necessary for some greater good (or for precluding some worse evil).

2. 2. There is at least one gratuitous evil, God’s permission of which is not necessary for some greater good (or for precluding some worse evil) (Adams Reference Adams1999: 82).

3. 3. So, God does not exist.

Skeptical theists can object to premise 2 (Perrine Reference Perrine, Edward and Nodelman2023, sec. 1.1–1.3). Just because we don’t see the greater good (or precluded worse evil) does not mean that it isn’t there. And given our cognitive limitations, we should expect not to see that greater good (or precluded evil). We are like children trying to figure out why their parents have allowed the doctor to administer a painful injection, unable to see the importance of the vaccine delivered thereby (cf. Morris Reference Morris2024; Wykstra Reference Wykstra1984).

It is tempting to think that the rapid development of artificial intelligence (AI) technologies can challenge skeptical theism’s appeal to human cognitive limitations. Notice premise 2 in the argument above relies on both our normative judgements about which events are good and evil and our causal judgements that some evil events are unrelated to greater goods (or do not prevent further evils). It is not so far-fetched to imagine an advanced AI system with a massive data set – a machine with a God’s-eye view of the world – helping us check whether the evils in fact causally contribute to a greater good (or screen off a worse evil). Skeptical theism therefore looks like a stopgap. Natural human cognitive limitations may be set in stone. But artificially supplemented cognitive limitations may not be.

This tempting conclusion should be resisted, however, for our imagined AI might be alien to us in the following sense. Machine learning algorithms notoriously track features of the information space that are opaque to us, such that although we can understand their outputs, we cannot understand the processes that generate them (Burrell Reference Burrell2016). We might have good reason to trust the AI, but we might be at a loss to understand why what it tells us is true. Indeed, we should expect not to understand the imagined God’s-eye AI given our cognitive limitations. Given this, our relationship to an alien AI is not so different from the relationship that some theists think we have with God. (I am one such theist.) We might have reason to trust in God’s goodness, and to therefore believe that there is a justifying reason for the permission of each evil, despite our inability to understand how any such reason is implemented with respect to any given specific evil (Stump Reference Stump2010: 11–15). Put otherwise, God is an alien intelligence, just like our imagined AI.

This similarity between God and AI bolsters the case for skeptical theism in the face of technological progress. In both cases, our only recourse may be to trust the process. But this comparison does not straightforwardly vindicate skeptical theism. Instead, it reveals what is truly difficult, and perhaps intractable, about the problem of evil. For, traditionally, God is a person. Even if we were to know by future technological means that every instance of evil causally contributes to a better world, we might still demand to know why this causal structure was chosen for creation. Why make a world where there is even the appearance of gratuitous evil? Perhaps we can live with the mysteries of AI, but a God worth believing in cannot remain alien forever.

These reflections on intelligence – artificial and divine – support Marilyn McCord Adams’ suggestion in Horrendous Evils and the Goodness of God; God must Himself defeat each evil, not merely outweigh them with greater goods, and perhaps this is only possible in an afterlife (Reference Adams1999). Even supposing our technological prowess reveals to us how each evil contributes to a greater good or precludes a worse evil, we would still lack a relational understanding that would allow us to meaningfully integrate God’s permission of evil into the story of our lives, a story where our suffering, or even committing evil, was not just a means to an end. Our technology, increasingly inscrutable itself, cannot help us develop a better relationship with an apparently inscrutable God.

Perhaps we will vicariously transcend our limitations in the creation of hyper-intelligent, genuinely conscious artificial beings who are up to answering the problem of evil in a way I cannot foresee. Still, I remain skeptical. Our post-human descendants, however impressive they may be, will not be able to think themselves out of the interpersonal challenge at the heart of the problem. When I try to envisage beings capable of knowing what I can only hope to be true, that the world is structured in all ways for the good, I see a necessary lacuna in their understanding. I cannot fathom their being able to understand why God left us, their forebears, in painful ignorance of the ultimate goodness of the world. I should like to think such beings would be indignant on our behalf.