Francis Turretin on AI

Yes, I will admit that title is more-than-a-little click-baity, but I didn’t get dozens of readers by playing it safe. And also, I’m not here trying to chime in on the recent controversy provoked by TGC’s AI stance; though it was occasioned by it. Nevertheless, the controversy did help coalesce a variety of thoughts, and as it also happens I am doing some research on angels for my 2 Peter commentary (2:4), and so…blog post.

Human Knowledge

Although there are people talking about AI from an epistemological perspective, they seem few and far between. The AI debate is ultimately about anthropology, and more specifically, epistemological anthropology. How do humans know things? Let’s not dive in too deep here (though if you want to, Frame’s DKG is a personal favorite). Suffice it to say, human knowledge is, by nature (and I mean that in every since of the term), embodied. That is, it is mediated through the senses and in some mysterious way “processed” and stored in the brain. Now it’s not only embodied (so don’t @ me Gray Sutanto), but our bodies are an essential component of how we learn, understand, remember, and know.

Angelic Intelligence

Which brings us to Francis Turretin. He wrote about AI before there was AI, though in his case AI stood for “Angelic Intelligence.” Like many scholastics before him (Reformed and otherwise), Turretin has an extensive discussion of angelogy in his Institutes. The questions he poses may seem speculative to us (they’re not, actually, but that’s a different post), but for him they are important, and he believes the answers are found in biblical exegesis and rational deduction.

One of those questions concerns angelic epistemology. How do angels know what they know? For Turretin, they have to know in a way fundamentally different than us humans. Why? Because angels are “spiritual and incorporeal” (I.541). They do not have bodies. They cannot see or hear or touch things like we can, so they cannot learn in the way humans learn; and, if we accept Turretin’s premise that angels are completely immaterial, we moderns can take this even further, since angels would then lack every human “instrument” for knowing, like a central nervous system.

What, then, is the “mode of angelic knowledge?” Turretin does not want to speculate too much, but he is bound by logic to affirm:

(1) that the angelic knowledge is only intellectual, not sensitive (the latter is organic and belongs only to bodies, which the angels have not). Hence no corporeal act is properly ascribed to them. Neither do they need corporeal organs or external senses which, as windows, they may open to objects. (2) That angels do not understand by their own essence; for thus the angelic essence would contain all things and be the cause and image of them (which belongs to the infinite God alone). Rather they understand by certain representations and ideas or similitudes of things, partly inborn and implanted from the first creation of things (which they know naturally), partly acquired (of things which they know either from divine revelation, God presenting such ideas to their intellect), or from experience (by representations abstracted from the things themselves). However although we do not know how they can abstract such representations or images from things (since they are destitute of external senses), we are not on this account to deny that this is done by them (I.543).

Artificial Intelligence

The point of all this is not to agree or disagree with Turretin. Rather, it is to recognize first (1) that he is concerned to ask this kind of question, (2) that he thinks about it carefully and through the lens of scripture, and most importantly (3) that he distinguishes different ways of creaturely knowing (he has similar discussions, for example, regarding the animals).

So now the contemporary question, which is to say another kind of intelligence, Artificial Intelligence. Turretin is not concerned about angelic knowing for its own sake; rather, it is because angels have interacted with humanity and are part of our story. He wants to know about them to better understand how we interact with them, and they with us.

That’s the question I think we should be asking about AI. Yes, there are serious ethical questions and practical matters, and we need to discuss those too, but we cannot really answer those questions without answering something more fundamental. How does AI “think?” It’s not an easy question, and if you think you know the answer to it than you’re smarter than most computer scientists because the one’s that aren’t in marketing will basically tell you that no one knows how generative AI really works.

Turretin has another profound insight. Embodied knowing is a different sort of knowing. There are many things that make human knowledge unique, the most important of which is a functions of the imago dei. Another important factor, though, is that God designed us with bodies, and our bodies are not accidental but essential, and further that our thinking and our bodies are tied together (in a way that we do not entirely understand). Our bodies are epistemologically important. The gnostics are wrong. The ancient gnostics were wrong, and the modern gnostics are also wrong. Human thinking is embodied.

Computers do not have bodies. Computers “think” differently than humans do, and that’s important, even and especially when the results of their thinking seem so similar to our own. And even that may not be true for long. Scientists and programmers know that the way AI works and learns is fundamentally different than the way humans work and learn. They also know that for many types of problems, humans are radically more efficient. Which is why the latest research involves “biological” computation (for an introduction, see here, but don’t listen right before bed).

Theological Resources

My point is pretty simple, I guess. We need to think theologically about this, and the resources for thinking theologically about this are already there in the tradition. The ethics of AI may be a new question, but it is not an unprecedented one. The tradition provides a trajectory for thinking about intelligences that are not human, and we can extend that tradition to start thinking about human intelligence as it interacts with other kinds of creaturely “intelligence.” Do you not know that we will judge angels?