Intelligence, Artificial Intelligence, and the Catholic View of Human Nature
by Joseph Vukov
Header image created on NightCafe by Michael Burns
In late 2022, ChatGPT showed up everywhere: in New York Times headlines; in YouTube videos; in theological conversations; in the policies of hand-wringing university administrators. In case you missed it, ChatGPT is an artificial intelligence (AI), a “chatbot” that produces text using machine-learning. Users can ask it to compose essays, write code, plan a child’s birthday party, or write a haiku. Or just have a conversation with it. The responses ChatGPT generates are not copy-pasted from somewhere on the internet, nor are they hastily penned by an English major in Cleveland hustling for rent money. The responses are instead genuinely novel pieces of text, generated on the fly using some of the most complex algorithms ever created.
I recently asked ChatGPT to summarize the Catholic Intellectual Tradition. Here’s what it composed:
The Catholic Intellectual Tradition (CIT) is a broad and rich tradition that encompasses the history of ideas and culture within the Catholic Church. It is a living tradition that continues to shape the intellectual life of the Church and engage with the broader world.
The CIT is rooted in the idea that reason and faith are not in opposition but complement each other. It seeks to integrate faith and reason, and to bring the insights of the Catholic faith to bear on a wide range of intellectual and cultural issues.
At the heart of the CIT is the belief that all truth is ultimately connected and that a Catholic understanding of truth can shed light on every aspect of human experience. The CIT has developed over centuries, drawing on the wisdom of many different disciplines, including philosophy, theology, literature, art, and science.
The Catholic Intellectual Tradition has contributed to a wide range of intellectual and cultural developments throughout history. It has played a role in the development of natural law theory, human rights, social justice, and the scientific method. It has also shaped literature, music, art, and architecture.
In the contemporary context, the CIT continues to be relevant to many issues facing the Church and the world. It provides a framework for engaging with questions of social justice, ecology, bioethics, and interfaith dialogue. It also offers a way of understanding the role of the Church in contemporary society and of articulating a Catholic vision for the common good.
Overall, the Catholic Intellectual Tradition is a rich and dynamic tradition that has played a significant role in shaping the intellectual and cultural history of the Catholic Church and continues to be a vital resource for engaging with contemporary challenges.
You know what? Not bad. A bit vanilla, and not much detail. If one of my undergraduate students turned in that essay, it could probably get a B (maybe a C). But tweak it a bit—sprinkle in some details and examples; personalize the prose; throw in a citation or three—and you’re well on your way to a decent essay.
The rise of AI doesn’t stop with text generation. For example, DreamStudio uses an algorithm—a very different one from ChatGPT—to generate images: faux photos; computerized Cubism; plausible post-Impressionism. I asked it to create a painting of a “Catholic on a computer” and it delivered Image 1.
Image 1: Created on Dreamstudio by the author
The result is not destined for the Louvre—and I’m not sure exactly what the hat is supposed to be (though it does look vaguely “Catholic”)—but the end result is not bad. Certainly better than anything I could create.
ChatGPT and DreamStudio are already two of many AIs that use machine learning. As machine learning becomes more powerful, and as creativity and ingenuity (and, let’s be honest, profiteering) spur Silicon Valley entrepreneurs to figure out ways to leverage the technology, ChatGPT and its cousins will soon be as ubiquitous as iPhones. And the AI that gets deployed will make ChatGPT look like child’s play.
Spoiler alert: the future is already here. ChatGPT is powered by a language model called GPT-3. The developers of ChatGPT, however, have recently launched GPT-4. And, as described in the interview with Blake Lemoine in this issue of Nexus, Google is currently working on an even more advanced language model called LaMDA, a model which attempts to integrate language with human personality. As Blake Lemoine puts it, the difference between ChatGPT and LaMDA is like the difference between appleseeds and apples.
A natural question arises as we interact with AI and reflect on what its future holds. Or at least, a natural question arises if you are trained as a philosopher, as I am: what is this thing with which we are interacting? AIs perform tasks we had thought only humans could perform. So…are these artificial intelligences human? Or—given a few decades and some extra polish—could they be?
When this question gets asked (whether in academic journals or sci-fi novels or drinks after work), we are often asked to choose between one of two sides. The first embraces AI as genuinely intelligent, poised to replicate fully the human experience. According to proponents of this position, if something walks like a duck and looks like a duck and quacks like a duck…it is probably a duck. Same with AI: if it chats like a human and composes like a human and creates like a human…it is probably a human. The second side rejects this way of thinking, arguing that AI lacks an extra “spark” essential to humanity and that no amount of coding or polish will ever produce something genuinely intelligent. There is, according to proponents of this way of thinking, some feature humans have—perhaps creativity or consciousness or sentience or being made in the Imago Dei—that AIs do not and cannot have. Humans are thus distinctive, and AIs can only mimic the real deal.
For Catholics, the second option may seem the obvious choice. For Catholics—and for plenty of others beside—the thing that is most important to being human cannot be captured by a programmer. Exactly what human distinctiveness consists in can be difficult to pin down. But whatever AI is capturing, well…it isn’t that.
I am sympathetic to this way of thinking. Yet I believe it is misguided to simply label the second choice as “the Catholic option.” Indeed, I believe both responses ultimately miss the target. The existential questions raised by AI point to deeper, under-scrutinized assumptions about what makes a human in the first place, questions that are often overlooked by all parties in the conversation.
The existential questions raised by AI point to deeper, under-scrutinized assumptions about what makes a human in the first place, questions that are often overlooked by all parties in the conversation.
To see what I am getting at, consider the following scenario: suppose you are an obvious Chicago native (Cubs hat; Chicago accent; currently ranting at the idea of ketchup on a hot dog). Being an out-of-towner, I approach you. “I’m looking for Greenwich Village in Manhattan,” I explain. “And I need to know: do I head North up Michigan Avenue, or turn East on Randolph?” “Well,” you respond, removing your Cubs hat to scratch your head, “I guess technically heading East on Randolph will get you closer to Greenwich Village. But neither option is really any good. The issue here is that you took a wrong turn a long ways back. You’re in Chicago, not New York City!” In short: while my problem superficially seemed to be which direction to turn, the real problem is that I took a (very) wrong turn, a long ways back.
Contemporary debates about the nature of AI likewise turn on assumptions that result from having made a wrong turn. According to both sides of the debate, intelligence is what characterizes humans most centrally. So, when artificial intelligence comes along, it seems crucial to decide: is this thing replicating or merely mimicking human intelligence? Does it capture or gloss over the human spark that makes us distinctive?
According to the Catholic view of human nature, however, focusing exclusively on questions of intelligence—and more generally, on questions about the human mind or consciousness—misses the mark. According to the Catholic view, humans are not first and foremost intelligences or minds or even conscious beings. We are instead rational animals, body-soul composites characterized by a capacity for reason. Our intellect—our rationality—is thus part of what distinguishes us from other animals. We can’t skip over the animal part to get to the rationality or intelligence part. Indeed, according to the Catholic view of human nature, the human soul—what makes us distinctive—is not something disembodied, but rather something that is embodied in a human way.
...according to the Catholic view of human nature, the human soul—what makes us distinctive—is not something disembodied, but rather something that is embodied in a human way.
The point is no mere footnote to the Catholic outlook. It is front and center. The idea that humans are essentially embodied shapes the Catholic understanding of resurrection. And of sacramentality. And of incarnational theology.
The upshot for our discussion? If we focus our debates about AI exclusively on the nature of intelligence, our debate is largely happening in the wrong place, much as our conversation about getting to Greenwich Village took place in the wrong city. Shift your discussion to the appropriate setting and debates about AI begin to look very different. It becomes less important to decide whether ChatGPT or DreamStudio (or its cousins) are conscious or sentient or psychological. Instead, we need simply ask whether an AI is (or could be) an embodied, rational animal in the way that Catholics understand human beings. Asked this way, the answer is obvious. An AI lacks a human body. So…not human. Case closed.
This doesn’t shut down conversation about AI, but rather opens new lines of inquiry. Could AI someday attain consciousness? Might it someday achieve a level of sophistication that it is worthy of moral consideration, perhaps in the way non-human animals are worthy of moral consideration? What kind of care should we take as we develop and deploy new forms of AI? These questions are both highly controversial and wide open. And they are questions that thoughtful Catholics should be front and center in formulating answers.
Joseph Vukov is an Associate Professor of Philosophy at Loyola University Chicago, where he is also an affiliate faculty member in Psychology and Catholic Studies. He is the author of Navigating Faith and Science (2022) and The Perils of Perfection: On the Limits and Possibilities of Human Enhancement (forthcoming 2023).