Interview: “The discourse around AI has quasi-religious overtones”

Dr. Christoph Heilig leads the “Focalization in Early Christian Stories” early-career research group at LMU’s Faculty of Protestant Theology and is a member of the Young Academy at the Bavarian Academy of Sciences and Humanities BAdW. In the “Theology between holy scriptures and AI-generated texts” research project at the academy, he investigates what effects large language models have on the hermeneutics and methodology of interpretation of the New Testament, and what contribution theology can make to AI discourse. In addition, Heilig teaches at the University of Basel as a Privatdozent.

Enlarge

As a theologian, what drew you to AI as a subject of investigation?

Christoph Heilig: The discourse around AI has long assumed quasi-religious overtones. People talk about a superintelligence, of omniscience and omnipotence. In theology, we have a wealth of experience in how such categories can be misused, but also in how to think about them constructively. So I think theology can be tremendously useful here.

To what extent does the idea of a superintelligence have religious overtones?

Naturally, this does not match the conception of God that we have in Judaism and Christianity, or in Islam, with God as the ultimate ground of all being who exists outside the universe. But there are countless other religious conceptions, such as those of Greco-Roman antiquity, where deities were responsible for certain domains which are very much part of the universe.

AI and the desire for immortality

Do you mean, people have this idea of AI as a sort of god?

Precisely. The superintelligence becomes a deity. And a coherent ethics and eschatology – the doctrine of final things – grow up around this conception of God. Eschatology is the decisive element here: The superintelligence will be able to conduct research – in the medical sphere, for instance – to develop rockets that procure resources from outer space, to program itself, and to build quantum computers. Many in the AI industry think this will lead to the possibility of digitalizing humans.

How would that work?

It’s about the attempt to combine technology and humans. Transhumanism used to be a niche ideology, but now it’s virulent in the AI industry. Humans must be transformed into a new being, so the theory goes. In the first step, by swapping out body parts and through brain implants. The end game is to download human consciousness on to a computer and transfer it to the cloud. Then we won’t need bodies anymore.

So a conception of immortality underlies the ideology?

Absolutely. One of the most influential health influencers in the anglophone sphere at present is Bryan Johnson, a millionaire who has made it his mission to become immortal. He’s become quite open about living a healthy lifestyle because super-AI is coming, which will be able to beam us into the cloud. For him, it’s about experiencing that and also proving himself worthy of it. The idea behind this is that AI could also wipe out humans. Because it is so clever, it knows what we humans do badly. In earlier times, we would have called these sanctification processes. We find this in the history of Christianity, where people have tried to present themselves to God in a better light. Through asceticism, for instance, or the sale of indulgences. Today people do this, say, by donating to firms in the AI industry.

Why does it interest you that individuals believe in this type of immortality?

I find it problematic – and not only at an individual level. There is the notion that we could turn the whole universe into one big supercomputer in this way, and could create all these digital spirit-beings who would be happy all of the time. When we combine this vision with a utilitarian ethic seeking the maximization of happiness, it follows that humans should spare no effort to bring about this superintelligence. At the latest, theology must interject when this potentially great happiness is purchased with the current suffering of (actually) living people.

This long-termist perspective should set off alarm bells for proponents of Protestant ethics, where individual responsibility of the human before God is so central that human life simply cannot be subordinated to such calculations. And if we are to make our actions so radically dependent on mere possibilities, such as to potentially affect millions and indeed billions of human lives, then we should be allowed to raise the possibility of the existence of God, in my opinion. And then perhaps other principles entirely would apply to ethical behavior.

What AI texts reveal about religion

We’ve spoken thus far about ideas that are circulating around the AI industry. But are the language models themselves and their text outputs changing anything about how religion is recounted and what is recounted?

Soon, I think, the so-called reasoning models will be so good that they will call into question existing religious conceptions and develop their own theology. At the moment, the direction of travel is still more so that of religion to AI output. It is abundantly clear that the large language models are not neutral as regards worldview. They were trained on the material that we all put online. In the case of the language models from America and Europe, these are currently Christian texts.

But the big global players, including those from China, all want to conquer the global market. So the developers are trying to ensure that the models do not ruffle feathers wherever possible. In doing so, they make an awful lot of decisions in advance. But this has engendered scant reflection in the AI industry, especially as such decisions evidently presuppose knowing what is right and proper for humanity as a whole.

Are you more troubled or fascinated by the current development of AI?

If you are not fascinated, then you have not been paying enough attention. With each passing month, so many things become possible that used to seem impossible. In the course of history, theology has often played a cautioning role, seeking to put a brake on progress because it did not recognize the benefits for people’s everyday life, emphasizing instead their spiritual salvation. But history has invariably proved that theology works best when it does not lose sight of the wellbeing of humans and draws on arguments rooted in creation theology that many good things are possible in this world.

Certainly it behooves us Christian theologians, whose religion is founded on the idea that “in the beginning was the Word,” to recognize, and not be all too surprised, that text can allow the world to be rediscovered and reshaped in fascinating ways.

For more on the research led by Christoph Heilig, see:

Website and blog