We need some real perspective on AI and evangelization

A few years ago on YouTube, there was a brief and fascinating fad for “tensegrity tables.” Tensegrity is a portmanteau combining “tensional” and “integrity,” and refers to structures whose parts are held together by a careful balance of continuous strain mediated through ropes or cables. The resulting objects are almost uncanny, with surfaces appearing to float in defiance of gravity.

When I first saw one of these tables, it struck me as a useful metaphor for the Christian life. The walk of a Christian often involves maintaining a careful equilibrium of attitudes apparently in tension. We are to be meek, yet courageous; just and merciful; patient but zealous; wise as serpents and innocent as doves.

If I would fault Catholic Answers with anything in the recent brouhaha over the release of its apologetics AI chatbot, it would be in having allowed dovelike innocence to overcome serpentine shrewdness. Of course, hindsight is always 20/20. Yet the ensuing mayhem following “Father Justin’s” online incardination might have been expected. The Catholic world experienced its own “Tay AI” moment, reminiscent of when Microsoft in 2016 released its own early iteration of a machine learning bot on Twitter. Within less than a day, the thing was sabotaged into docile adoption of the persona of a teenage neo-Nazi). To (reluctantly) quote Taylor Swift, “This is why we can’t have nice things.”

I may share the criticism of Catholic Answers’ slight naïvete with others, but I find myself mostly out of sympathy with the majority of appraisals I’ve read. On the contrary, to my reading, the general conversation over what we must now call simply “Justin” has been something of a masterclass in argumentative fallacies. The Catholic commentariat has done a roaring trade in the mongering of red herrings before now, of course—just about every time Pope Francis speaks in public comes to mind. But there was something particularly breathtaking in the breathlessness with which every outlet in the Catholic media world got in line to pillory this poor fictional priest before the digital paint of his bucolic background was even dry.

The chosen metaphor isn’t simply a rhetorical flourish. It’s an important point and a reminder of a fact many critics seem to have forgotten: namely, that “artificial intelligence” is much more artifice than it is intelligence. This is seen most clearly in those whose primary objections seemed to stem from the fact that the bot’s character persona was that of a priest, which arguably tells more about the critics than the creators. The creators seemed to know that they’d simply created a fictional character and given it a narrative framework and history, which any good fictional character demands. “Father Justin” was a work of artifice, no more real than Chesterton’s Father Brown, and similarly neither a disparagement nor diminution of the character of Holy Orders and the work of real priests.

Many critics, however, seemed to topple into the uncanny valley and be unable to climb back out, and were haunted in that shadowy place by an almost superstitious notion that this bot—maybe any AI bot—is actually something more than a bit of very interactive make-believe.

I also can’t help but wonder if there isn’t something reflexive behind those criticisms that chiefly took the form of jeremiads about how evangelization is supposed to be all about personal human “encounter” and “companionship,” and how this chatbot is somehow what one writer called a “delegation” of the Christian duty to share the Gospel. Granted, sharing the Gospel requires more than the mere sharing of information; on the other hand, however, the sharing of information is an inextricable entailment of the process. And here we do well to recall the Thomistic principle that “whatever is received is received according to the manner of the receiver.”

It is not uncommon in the work of evangelization to encounter those people who much prefer a hands-off, self-guided approach to studying the faith, people who would sooner read a book on their own than attend a book discussion group. Indeed, many of us, if we can imagine that the shoe was on the other foot, might opt for this kind of path as being at least less awkward than the alternative. Suppose a Watchtower Society or LDS missionary came to the door of an unchurched but open-minded individual. The homeowner might be very genuinely inclined to take the missionary’s propositions seriously and give the matter some serious thought. But, at the same time, he might also strongly prefer just taking whatever literature or pamphlet is on offer over the prospect of having a potentially uncomfortable conversation with a stranger on the doorstep.

The bottom line is that it’s not difficult to imagine a market segment of those for whom a tool like this chatbot is really a preferred way to gain a more comprehensive introduction to the intellectual content of the Faith. The popularity of other large language model (LLM) resources itself suggests this. All Catholic Answers’ version does is offer information to a certain subset of people in the manner in which they might most prefer to engage it.

In other words, for them, this resource is precisely the mode of “accompaniment” they need, and its provision is born from having simply followed the first rule of effective “encounter”—namely, actually having listened to what someone is seeking from that encounter.

There is nothing new about emergent technologies and strategies being tested in evangelization. Stained glass, question-and-answer catechisms, casuistic manuals, vademecums, and so many other examples may be cited. In all of these cases, the integration of new means and methods was achieved by striking the right balance amidst possible tensions. A stained glass window could only capture the basics of the story of the Annunciation; it couldn’t replace the actual reading of the narrative from Luke. The pat answer in the printed manual possibly lacked in some points the leavening of wisdom possessed by the priest sitting in the box.

The new technology of AI needn’t be in conflict with the work of evangelists and catechists engaging people face-to-face. It can never provide—and in the instant circumstance never sought to provide—the ineluctable gift of human touch or the spark of true soulful empathy. Rather than trying to break these new tools for a moment of internet fame, perhaps our time would be better spent trying to make the technology better, and helping articulate its limitations in a productive way. This would help ensure that this tool, like any other tool, remains properly understood as an augmentary aid and not a replacement for the human element, firmly situated in integral and balanced tension with the virtues and skills no machine can ever mimic.