What ChatGPT reveals about our history—and demands of our present
Tell me, in the style of Jesuit priest Sam Sawyer, about what ChatGPT means for religion:
Well, my friend, the emergence of ChatGPT certainly presents an interesting opportunity for the intersection of technology and religion. As a Jesuit priest, I believe that faith and reason are intimately connected, and that our ability to understand and communicate with one another is a fundamental aspect of our human nature.
In that sense, ChatGPT can be seen as a tool for enhancing our ability to engage with religious traditions and ideas in new and dynamic ways. With its vast database of information and sophisticated language processing capabilities, ChatGPT can help us explore complex theological concepts, engage in interfaith dialogue and even offer guidance and support to those seeking spiritual direction.
Of course, like any technology, ChatGPT is not a panacea. It cannot replace the human connection that lies at the heart of religious experience, nor can it provide the kind of personal relationship with a divine being that many believers seek. However, I believe that when used wisely and in the service of the greater good, ChatGPT has the potential to enrich our religious lives and help us better understand the mysteries of the universe and our place within it.
You need not worry that I will be outsourcing “Of Many Things” columns to our robot overlords in the future.
That’s…not too bad, actually. I don’t think I’d really begin with “Well, my friend” but the rest of it is an embarrassingly good mimicry of my prose style; presumably the America website was in the corpus of internet text on which the model was trained. (Though it has not picked up on our house style; there was an Oxford comma in the prose it generated before we copy-edited it.) And I definitely do believe that faith and reason are intimately connected and that the ability to communicate is fundamental to human nature. But I would never, for a host of reasons, suggest that ChatGPT could “offer guidance and support to those seeking spiritual direction.” Among the reasons, which is something ChatGPT got right, is that it certainly cannot replace human connection or a personal relationship with God.
You need not worry that I will be outsourcing “Of Many Things” columns to our robot overlords in the future. The start of this column was actually inspired by a colleague, Zac Davis, who gave ChatGPT a prompt to “write in the voice of Jesuit priest Sam Sawyer” as a joke in the midst of discussions about how these new tools might affect our editorial work.
Before I was a Jesuit, I was a software engineer. I never worked in artificial intelligence, but I followed developments in the field and I knew the standard joke about how A.I. had been just 10 years away for the past 40 years. The large language model chatbots and generative A.I. image tools are the first developments I have seen that really seem to disrupt that always-off-in-the-future timeline.
These new A.I. tools can profoundly accelerate some kinds of understanding in ways that will be both helpful and harmful, as all tools can do.
What is perhaps most disconcerting about the new A.I. chatbots is how much talking to one feels like having a “real” conversation, even when you know that in the background there is just a giant model predicting what word should come next in the chain. By the time you read this column, America will have published on our website a selection from a conversation about A.I. between two Loyola University Chicago professors and Blake LeMoine, the former Google A.I. researcher who made news last summer over his concern that an A.I. called “LaMDA” had become sentient. In the excerpt from the journal Nexus: Conversations on the Catholic Intellectual Tradition,he argues that when we try to get an A.I. to understand humans, it ends up, as he says, “somehow, some way, perhaps metaphorically,” experiencing what humans experience.
One reason talking to an A.I. feels like a real conversation is that we are talking, through elaborate mediation, to all of the human text and learning and history on which the A.I. itself was trained. We are in that sense—somehow, some way, perhaps metaphorically—talking to ourselves, each other and our tradition.
And that can be discomfiting because it confronts us with how much of our own experience of one another is mediated by language and conversation, how much we are ourselves modeling the experience of others based on the words we exchange with them.
But—and here’s why I won’t be sending anyone to ChatGPT for spiritual direction—while we are inevitably modeling other humans in order to understand them, our relationships are not fulfilled solely in understanding, but in loving. It is charity and willing the good of the other, not mere comprehension, that most deeply undergirds human connection. These new A.I. tools can profoundly accelerate some kinds of understanding in ways that will be both helpful and harmful, as all tools can do. The task of turning them to the works of love, however, will remain ours.