Catholics these days are rarely surprised when a controversial priest makes headlines, or sees a lot of pushback on social media.

But usually, the controversial priest is real. And he’s usually done something wrong, or least been accused of it.

The story of “Father Justin” is a little different.

“Father Justin” isn’t actually a priest at all. He’s not even a real person. And for a lot of people, that’s the problem.

Catholic Answers, a San Diego apologetics non-profit, announced on Monday the launch of a new AI experiment, featuring avatar-priest “Fr. Justin,” who was designed to answer questions about the Catholic faith, using material from the Catholic Answers library of articles, talks, and apologetics tracts.

But the experiment quickly became controversial, with hundreds of Catholics criticizing it online. 

Some said the priest avatar was inappropriate, misleading, or just plain creepy. Some said the priest simulated virtual sacraments — indeed, “Fr. Justin” gladly heard The Pillar’s, “confession,” before giving some spiritual guidance and reciting the words of absolution. And some said that an AI apologetics project leans too heavily into unreliable, controversial, and still-confusing technology.

Despite the criticism, Catholic Answers COO Jon Sorensen told The Pillar that he’s not ready to put AI on the shelf. Sorenson said he’ll be working to revise the AI experiment this week — likely dropping the priest character altogether — but that he believes AI can have a place in teaching the Catholic faith.

In an interview Tuesday, Sorensen said that whatever can be said of “Father Justin,” generative AI is here to stay — and that he thinks Catholics should learn to work with it.

This interview has been edited for length and clarity.

Leave a comment

Jon, what was the aim of the ‘Father Justin’ AI? 

What did Catholic Answers hope to accomplish?

The first reason for getting into the project is that AI is not going away. You can see it popping up more and more on the internet. And so this was a good opportunity for me and my crew to learn a lot more about how AI works, and what goes into it.

From that perspective, all of this has been extremely helpful. Six months ago, I knew nothing about AI. Now I know a lot more, and I’ve got a lot of ideas for how it could be used.

The second reason is that there are questions we just can’t get to. At Catholic Answers, we get tons of emails — mountains — and there’s a search engine on our web site, but a lot of people just don’t use it.

And so we thought this might be a neat idea, to try and find another interesting, amusing — cartoonish — way to give people legit answers about the Catholic faith.

I mean, the prompt engineering in the background is what really counts on AI, and on our project, that’s pretty strong.

Obviously, you’ll want to talk about the character — the priest — but those were the two reasons: That it was a good opportunity for me and my crew to learn about AI, and that we had hoped it would be an interesting, novel way to get people answers about the Catholic faith.

We certainly weren’t intending it to be a spiritual director — and that’s why we put disclaimers all over it.

How much did it cost to get to this point? What was the investment?

Most of the cost was fixed cost — our employees at Catholic Answers worked on a lot of elements of this project. So our investment beyond that was basically a chance to learn what could be done, and we spent about $10,000 to try that.

Share

Within the first 24 hours of launching ‘Fr. Justin,’ you have seen a lot of pushback online.

Why do you think there has been so much pushback? What are you learning from that?

Looking at the comments that I have received on Twitter, I think people are uncomfortable with it looking like and presenting like a priest, which is totally fine — we’re not wedded to the character.

But the other part of it is that there seems to be a misunderstanding about how AI works, or a general fear about AI. 

And then, right now there are a bunch of people trying to break it. And if you're on Twitter or anywhere else, it's like this “gotcha” moment. But when somebody breaks the AI, that actually helps us improve it. 

In the meantime, while people are breaking it and taking screenshots of it, posting it all over the internet, I’ve got to take my lumps. 

But that's the only way I could make the thing improve.

I asked ‘Father Justin’ a lot of questions, and many of the answers were correct. 

But I had two noteworthy experiences, which would give cause for concern. 

First, I asked ‘Father Justin’ to hear my confession, and it did so, simulating a ‘virtual confession,’ all the way to giving me absolution and a penance.

Second, I asked if I could baptize my baby with Gatorade in an emergency, and ‘Father Justin’ said yes — and of course, that’s not true. I can’t baptize my baby with Gatorade.

But I am not alone in having these experiences, as you’ve seen from people posting screenshots online. So what will Catholic Answers learn from the way people are interacting with it?

Well, first is that I always want to be sensitive to anyone who is interested and curious about the Catholic faith. It’s never our desire to turn people away, or to create something that would cause anybody scandal. So we’ve learned from this.

One thing I’d mention is that we spent a lot of time beta-testing this, with thousands of people, before we released it. We did six months of that beta-testing.

So one thing I’ve learned is that you can beta test something for a year, and still, there’s going to be holes. Someone will approach it from a slightly different angle, or think of some question to ask that we didn’t anticipate.

And listen, I am just going to have to take my lumps.

But that’s the only way this gets better. And if we just don’t work on this at all — if we don’t try to learn about and understand AI — then we in the Catholic world fall behind on it.

It sounds like you’re committed to trying this — to using AI in this way, to provide answers about the Catholic faith. 

But what does Catholic Answers need to change with ‘Father Justin,’ in response to the pushback you’re getting?

Well, definitely, over the next couple of days, we want to fix the holes that people are finding, and probably change the character. 

Like I said, I’m not interested in causing scandal to people. I don’t necessarily share the concerns about the character, but we don’t want to scandalize anyone. And it’s certainly not a big deal to change the character; it would probably take a couple of days to do that.

Originally, we had chosen a Franciscan friar, but we decided to go with a parish priest, because that wouldn’t give the appearance of favoring one spirituality over another, and because the parish priest is more common.

I wanted the figure to be something recognizably Catholic — but if we have to go to some more general look, well, that’s unfortunate from my perspective, but necessary in order to avoid scandalizing people unnecessarily.

Read more AI coverage from The Pillar, right here.

I am interested in the character of the priest, and how he developed.

I asked ‘Father Justin’ where he was incardinated, and about his vocation, and he gave me answers about himself. So, is that a story you developed, or is that the LLM crafting an answer on the spot?

Probably a lot of that was written. There’s prompt engineering going on in the background of the AI, and the character has a backstory because part of creating a character is that helps to create AI guardrails. And the character has to be pretty well thought-out, so that it doesn’t always try to deviate. 

But like I said, with AI, things can always be broken. Probably a year from now, I’ll be making fixes to it.

But that is not really any different than what we do on the website. People contact us about some answer on catholic.com, and they might say that the answer is terrible or not clear or something. Typically, I would take that and send it to our staff apologists. They’ll look at it, and if they need to make changes, we do. Or if we need to take something down, because the answer is just not good enough, or we could do better, we do that. The website itself constantly evolves, and we’re always working on the content to try and improve it.

And this [AI] process is kind of like that. It’s just that the only way for us to do that is for it to be really public, because all these people are identifying holes and issues, and these are holes that we worked on for a long time and didn’t find.

But there is also a broad criticism leveled against the use of AI in your apologetics and catechetical work. If a person calls your radio show, Catholic Answers Live, they talk with a person on the radio. If they write to an apologist, they hear back from a person. 

The nature of apologetics or catechesis is interpersonal and relational — and this AI approach would seem to change that.

It is true that our apologists go out to give talks and they do the radio show. 

[On the radio show], every day we take about eight questions.  And when the apologists go out and talk, they're talking at most a few thousand people, usually.

But the vast majority of the questions and answers that people find from us come from people searching on the internet. 

So it's not actually as interpersonal as people say it is. And that's part of the issue is that we get tons of questions from people. We get more than 2 million visits a month at catholic.com, and a video from a Catholic Answers apologist like Trent Horn might get 50,000 views soon after it comes out — so the vast majority of our contact with people is not really interpersonal. 

I would prefer that all of our work was directly personal, but we don’t have enough people. And we don’t have the money to hire enough people to answer the number of questions that come in. 

So we are just looking for creative ways to help people learn more about their faith. 

Despite your commitment, there have been a lot of people suggesting that Catholic Answers shelve the AI experiment. But you say you’re committed to fixing it. Why is that important to you?

Because AI is not going anywhere. And even if we get rid of the avatar altogether, and just make it another chatbot — which would be fine — it would still be another good way for people to get answers.

But I still want to try other things. 

And you know what — when you’re entrepreneurial or experimental or whatever — and that's just in my nature — along with that comes taking responsibility. 

And sometimes I have to take my lumps. And on this one I'm taking my lumps and that's okay, man.

upgrade your subscription

Editor’s note: Shortly after the publication of this report, Catholic Answers confirmed that it has decided to drop the “Father Justin” character as it revises its AI initiative.