Pastors, AI Is Here: 3 Questions You Should Be Asking

I frequently use generative AI for pastoral ministry.

• I use ChatGPT to craft emails, bulletin announcements, and other church communication.
• I use ChatGPT to create discussion or application questions for sermons.
• I use Logos' AI tool to summarize search results or to provide a synopsis of specific resources that I use during my study process.
• I currently create my own sermon slides and templates but wouldn't be opposed to using generative AI to do that.

Generative AI is a powerful tool, but you must use it wisely and responsibly.

I had a good chat with my older son this morning about AI. He'd had a chat with his girlfriend's father about the matter, and the father is persuaded that there are a lot of things for which AI will be increasingly valuable in the future. The father is currently using it for a number of repetitive tasks, which is a use which I can commend.

That noted, another point of agreement between the father and myself is that there are certain weaknesses that result from the over-use of any tool. For example, the use of calculators to do arithmetic has resulted in generations of people who can't do basic calculations without one.

In terms of AI, the same applies. Yes, it can generate simple memos--at the risk of never really learning to write. Use it to solve algebra and such? Same basic idea.

(how do you catch it? You force the student to write out "his work", and they periodically make mistakes in the middle of the transcription, but rarely at the end--so they get the right answer but in an impossible way)

But probably more important is that so far, AI works with the database it has, and hence if that database doesn't contain an idea, it's hard for it to extrapolate there. Already, professors and teachers are using AI detectors that attempt to discern whether a "student's" writing is their own, or whether its correlation to the AI database and output is strong enough to suggest cheating.

So if one's line of work depends on actual design rather than averaging out what's already out there, the current AI state does not yet pass the Turing test.