To Use or Not To Use ChatGPT

There Is Really Only One Question

Cutting through the noise to get to the bottom of the pile.

First, Kill Your Naivitee: It’s Happening All Around

This year, my undergraduate Christian Ethics class was awash with Chat GPT use. The irony of using an LLM to game a Christian ethics course aside, I get it: it’s a general education course and almost none of my students had a deep investment in—or knowledge of—what ethics was before the class.

I went into the class assuming that, though some were disinterested, they were at least giving their best shot. But in running all of their midterm essays through the embedded AI detector in Canvas, the detector flagged seven papers which were more than 50% constructed using a LLM of some description. I had already included a provision in my syallabus prohibiting their use, but mostly as a defensive measure, a kind of CYA proposition.

The discussions with these seven students was fairly straightforward about the reduction in score. They got it, even if they had used it varying degrees: some for organizing thoughts, some for straight up generating the whole paper. What was clear in the syllabus was that it was forbidden.

What was far less clear to them was why.

Many Questions, and The Main Question

As an ethicist, there’s any number of questions which artificial intelligence generates. Like you, I’d love to know whether something that resembles humans but is a bunch of 1s and 0s has a soul¹. Like you, I speculate as to whether LLMs reveal that my writing is just inhuman, given that a machine with enough time can emulate my voice. Like a very small number of you, I wonder whether electric sheep dream? And like at least two of you, I wonder whether Shakespeare’s genius can be so broken down by machine learning that we get a new sonnet or tragedy from the Bard?

But, as I’m an educator, I’m going to focus my comments more specifically: on the intersection of the use of LLMs—Large Language Models like ChatGPT— and Christian moral reflection.

In asking about the ethics of AI and its related uses, there are any number of approaches we could take: academic integrity, violation of copyright law, ecological impacts of server centers, and the like. But these all seem to be downstream from more basic concerns, which is in what way do the myriad forms of AI comport to a Christian vision of what it means to be human?

There is no direct command in Scripture governing their use, and certainly no form of pro/con analysis that provides clarity on this. Humans are notoriously good at finding ways at letting efficiency be the unspoken metric of our moral deliberations. But the kinds of things which God commands, and the kinds of pro/con lists which we can make all ultimately find their locus in this primary question of what it means to be a flourishing and faithful human.

It is, in other words, a question of virtue.

What Does It Mean to Be Virtuous?

This is a difficult question, but for starters, virtue means something other than being nice. Being a virtuous person means, to be sure, being a good member of a society: it’s hard to be nice without being nice toward someone.

But virtue has to do with how one’s character resonates with the grain of the universe, with what one is meant to be as God’s creature. There are theological dimensions to this: faith, hope, and love. But this is the culmination of virtue, what it means for the creature to be restored in Christ. All persons, by virtue of having been created by God, are meant to have a certain kind of character, a kind of character which speaks to the order God intends for diverse societies.

Historically, this substrate of human excellence is called the cardinal virtues, of which there are four, umbrella virtues which provide orientation for all other virtues: fortitude, prudence, temperance, and justice.

We can think of them, roughly, like this:

Prudence—knowing how and when to do a thing, neither too early or too late.

Temperance—knowing how much of an action is needed, neither too much nor too little.

Fortitude—continuing to do good things in the face of difficulty

Justice—how good things flourish well among people.

These four together comprise what we name as “wisdom”: knowing not only what is good, but how to develop judgment about how to do hard things, how to keep doing them, and how to share good things well. It’s a tall order, but if virtue isn’t just being privately nice, but being a bearer of the way things should be in God’s creation, then virtue is a fairly complex pursuit. The aim of all persons, then, is to pursue this kind of life, that the whole of creation might together bear witness to this way of God’s world.

To be virtuous is to have one’s character shaped in such a way that what we are—both outwardly in our actions and inwardly in our deliberation—is governed by a life directed toward The Good, or more specifically, God. Now, there’s all sorts of things to add here: what it means to be good isn’t the same as being happy², or even something as simple as it being pleasurable. Given that we seek God in the midst of a world set up to be indifferent to it, or at times hostile toward such aims, you have to have something else in view beside happiness.

All of the more familiar virtues—patience, courage, thrift, etc.—are all components of these bigger virtues, and help give texture to the cardinal virtues. So, for example, to be a person who keeps going on in hard times (fortitude), you have to be patient, courageous, encourage others, give honor to those who have succeeded, and so forth. All of the little pieces work into one giant cornerstone, or cardinal, virtues.

Virtue—what we are meant to be as humans—helps us to make sense of biblical commands, for God’s commands cohere to what it means for us to God’s creatures. The ultimate aim of God’s commands are that we be God’s people, and the virtues help us to name what it looks like for God’s wisdom to encompass the world, and what it means for us to move well through it. We are patient not because it is commanded, but because God’s articulation matches what God has created us to be.

Dinking the Human: LLMs and the Cardinal Virtues

With this in view, I want to briefly offer an analysis of LLMs through the four cardinal virtues. The list of these four varies slightly from thinker to thinker, but Thomas Aquinas charts them as prudence, temperance, fortitude, and justice. In looking at LLMs through this lens, we see the deeper stakes: whether this undermines our ability to pursue being human.

I confess that I am skeptical that AI, and particularly the LLMs which my students use, fit this bill. The use of LLMs undermines their ability to know how to make judgments about material and organizational patterns, failing the prudence test. Not being able to have a knack, to develop an intuitive sense of how things fit together or what information is relevant and how—this is the sign that we lack prudence, lack a sense of nuance needed to navigate the world.

Given the ease with which these make writing and design assignments, it’s hard to imagine that they will encourage students to be moderate in their use, or to cultivate a broader sense of temperance. Giving an easy way forward for all assignments, and having no sense of when AI might be augmenting our work—like Grammerly—and when it is doing all the lifting: this is a sign temperance is non-existent.

The difficulty of learning skills, when made more expeditious by the LLM, certainly undermines their ability to persevere in struggle, knocking fortitude off the list. Learning should involve struggle, because in struggle, the bad habits are unlearned, ignorance is undone, and insight becomes ingrained in us. The knowledge becomes a part of us, the skills intrinsic to our way of moving through the world.

As for justice, while it certainly may be fair to let all students make use of a widely available tool, it does not apportion to the student what they need to flourish intellectually or morally as God’s creature, and thus, is not just. As I put it to my students, the problem is less that they violated academic integrity than the fact that they robbed themselves of what was rightly theirs in the educational process.

The Mostly Happy Ending: Virtue Can Win

In light of the use of LLMs at the midterm, I scrapped the written final, and converted it to an oral final. I administered 10 possible questions in advance, let students prepare, and then we spent twenty minutes talking about the answers. The results were fantastic: those who had developed the capacity for long-term thought, who had deeply understood the material, and who could make connections between discrete topics did well. Those who had not put in the work did not.

One student who had used an LLM to generate 75% of her midterm did exceptionally well in the final, and it proved to be a truly pastoral moment to affirm for that student that a) she didn’t need the LLM to do her argumentation and writing for her, and b) the road of learning—the road of virtue—is slower, but ultimately more rewarding, as demonstrated by this final.

She saw the value in struggle, the ways in which she had learned to grow in being able to make arguments, develop judgments, and saw that her use of the tool—in my class and others—was immoderate and depriving her of the goods of an education. This is not yet virtue, for I am not sure that she liked the hard work, but to paraphrase Augustine, sometimes the semblance of virtue is enough until it becomes something you see as desirable for its own sake.

1. J/K. They do not.

2. Lots of virtue ethics would argue with this statement, particularly the granddaddy of the tradition, Aristotle. But it seems irrefutable to me that have a life comported toward God will mostly make you contented and seated,