The Social Imaginaries of AI

The Implied Worldviews of Silicon Valley and Leo XIV

I know, another post on AI. It seems like that is all there is these days. Concerns about labor and the environment, commentary on Magnifica Humanitas or expressions of anxiety about teaching (especially as the semester gets going), I’ve read about all of these. I might even have written on all of them. However, as the pope notes, “any statement regarding AI risks becoming quickly outdated, given the remarkable pace at which these systems are developing” (no. 98). So, I continue to think about it.

And I think what has struck me lately, especially trying to think through the theological implications of AI, is that we have multiple social imaginaries of AI. Borrowing (and simplifying) the term from Charles Taylor, I’m thinking about the way people – or in this case AI companies – imagine social existence, including how we relate to one another, what holds society together, and how common life ought to be ordered.

I think there are currently three main social imaginaries competing in Silicon Valley.

Peter Thiel and Palantir represent a broadly libertarian perspective. His concern is not primarily with AI but political institutions becoming too powerful. His recent focus on the anti-Christ is his concern about an emerging political order that suppresses AI as part of a promise of peace and safety but is really about just aggregating power. These anti-Christs range from Greta Thunberg to the pope. As a libertarian, this view does seem rooted in a concern for freedom, a belief that more freedom solves more problems and makes things better, and, thus, a suspicion of institutions because they tend to curb this freedom. As Thiel is a bit reluctant to believe that all human life is valuable, his social imaginary is basically a world where people, especially AI companies, can and should do whatever they want. For those who embrace this freedom, their lives will be great. Those who don’t, well, they are just stumbling blocks.

Image via www.vpnsrus.com at Wikimedia Creative Commons

OpenAI shares some of the wariness about government control and intervention, but it seems to have a strong sense of norms and ethics, at least ones that develop over time. It assumes that society learns through experimentation. The future cannot be fully anticipated, so the way forward is to act, reflect, revise, and redeploy, or as they say “pilot new approaches, measure what works, and iterate quickly.” Through this process, things become better. This is why OpenAI seems wary of initial regulations. They work better “downstream.” The issues and problems cannot be seen beforehand, so the way to solve problems and make things better is through iterative deployment. As I see it, their social imaginary reflects American pragmatism. Maybe not quite “move fast and break things” but a more tempered “move fast even though you’ll break things.” As an Appalachian, pragmatism appeals to me, but also as an Appalachian, I am wary of companies that are too pragmatic. The pragmatism of coal companies overrode any concern for people, the miners, their families, the community, the environment, the whole region.

Anthropic takes a different approach. It wrote a “constitution” for its AI, embedding it into Claude. As they explain,

Claude’s constitution . . . . both expresses and shapes who Claude is. It contains detailed explanations of the values we would like Claude to embody and the reasons why. In it, we explain what we think it means for Claude to be helpful while remaining broadly safe, ethical, and compliant with our guidelines. The constitution gives Claude information about its situation and offers advice for how to deal with difficult situations and tradeoffs, like balancing honesty with compassion and the protection of sensitive information. Although it might sound surprising, the constitution is written primarily for Claude. It is intended to give Claude the knowledge and understanding it needs to act well in the world.

This is different from either Thiel’s “freedom as the fundamental value” or OpenAI’s “we’ll figure it out as we go.” It is a commitment to principles that guide action. I’ve seen it. Whereas ChatGPT will affirm my expressions of frustration even when they are laced with expletives, Claude will simply reply, “I will not engage in this kind of behavior.”

This idea of a priori limitations on power is why its social imaginary is like a constitutional democracy that constrains power based on rights and competing claims. This is also closer to the Catholic faith, with its natural law tradition that is behind its commitment to representative government (a view developed by John Courtney Murray and Jacques Maritain). It is why, I think, Pope Leo had Christopher Olah, the co-founder of Anthropic, at the Vatican for the release of Magnifica Humanitas.

Even so, I don’t want to dub their view the “Catholic” view. Just as democracies and economies have problems because they cannot account for the whole human person, so too is Anthropic’s account limited. Its constitution is written for its AI, not humanity, so it will function well when it is AI doing AI stuff, like economics doing economic stuff and politics doing political stuff. But it will struggle with a full humanity, like that described in Magnifica Humanitas, no. 50:

At the heart of the Christian understanding of the human person lies the great biblical affirmation that men and women are created in the image and likeness (cf. Gen 1:26-27) of the Triune God. Created for relationship, every human person is planned and willed by God to enter into communion with him, with others and with creation. Human dignity does not depend on a person’s abilities, wealth or position in life, nor on the right or wrong choices made; instead, it is a gift that precedes and transcends each person, endowed by God as an expression of his unfailing love. For this reason, the human person always remains the “way for the Church” and the heart of every authentic path of integral human development.

Lifting up a view of the human person and making it the criterion for evaluating AI is, I believe, one of the great contributions of the pope’s encyclical. It highlights that one of the most significant issues we are facing with AI is their social imaginaries. This is a properly theological point to make, and it leads to ethical and social concerns. For social imaginaries will flow into relationships, the economy, politics, the environment, and everything, and if they aren’t right – or at least right enough – they will corrode and degrade everyone and everything, even if they are not constrained by governments, eventually get better, or are limited by their own internal values.