This is something that’s been sloshing around in my noggin’ for a while now.
How should The Church react to the current aspects of Artificial Intelligence (“A.I.”) and machine learning with Large Language Models (“LLM”)? I’m thinking along the lines of GPT-4, Bard, and creative image diffusion generators like DALL-E 2, Stable Diffusion, Midjourney, and the like?
I’m not talking about using the LLMs to write full-on sermons (though using for some illustrations might be interesting). I’m wondering about other areas that may be helpful.
Let’s be clear to start with, however: this isn’t going away. It isn’t a fad that will up and disappear. (Though some of the current uses are “gimmicky” in their infant iterations.) The fact is, The Church, like society, will need to wrestle with when, how, why, and even in what settings to use these tools. Frankly, this is true of all new tools and technologies. Even so, The Church always seems to lag at any implementation. (Just look at some church websites, which still seem to be stuck in the 90’s era of web construction, as if that’s a fad!)
The truth of the matter is that, like most anything else, this is a tool which can be utilized. It’s a tool that can be abused. It carries no intrinsic morality of itself. Implying otherwise anthropomorphizes that which carries no “imago Dei.”
I think it’s easy to consider some of the poor, non-beneficial, and even dangerous ways the technology can (and so will by some) be used. But what are some ways The Church could utilize this technology for the benefit of people and to the glory of God?
Some possibilities come to mind:
- Use good theological works, commentaries, Bible dictionaries, lexicons and the like to create a specialized LLM, so people can ask simple questions of complex topics and receive a good, logical, understandable explanation instead of just list of links or resources. e.g., “What is infralapsarianism and how should Lutherans view it?” “Do Lutherans believe you ‘sleep’ with you die until the second coming or do you immediately go to heaven or hell?” “What’s the difference in American Lutheran churches?”
Certainly there are many good, clear articles and publications for some of these, but probably not for all questions someone has or with the nuanced query that may be plaguing them. Perhaps we would find people asking good questions (or follow-ups) they didn’t know they originally had – or are hesitant to ask a pastor. This can especially be true for those who may not be in church or have had bad experiences with The Church. As more questions are asked and the generated answers tweaked over time and fed back into the model, a very good addendum to our theological corpus will arise.
Along the same lines, what if we fed words from good hymnody into a model and paired it with meter from more contemporary music, asking to generate some good doctrinal possibilities where hymns may not always work? - Feed the thousands of found manuscripts we have (including the confidence level for each, challenges, etc.) into A.I. to see if there are some interesting language pattern changes over time or by manuscript type, etc. It might be interesting to see if there’s something A.I. sees over those thousands that we haven’t (yet). With morphology feed in, perhaps we could ask more natural queries of the original texts along the lines of, “Show me the differences in the manuscripts that have John 7:53-8:11 and those that don’t. What are some other differences in other verses with those same manuscripts?” “Show me all the places where ‘love’ is the verb and ‘YHWH’ is the object of the sentence? What about ones where ‘YHWH’ is the speaker and ‘love’ is the verb?”
- When dealing with these tools, crafting the prompt and query takes on more of a balance between art and science. It’s not just what you ask, but how. This is especially true for the graphic models. How interesting would it be to see some A.I. interpretations of Biblical items given the Biblical descriptions? e.g., the Ark of the Covenant; the priestly garments of Aaron and his sons; and the Temple of Solomon?
It is true that several of these things can be (or have been) accomplished in other, probably better ways. But that may not always be the case. The fact is, these tools (and others that are yet to come) are here to stay. And The Church should be ready for it. Even better, it should find intrepid ways to use it for the Kingdom.