For better or worse—and there is plenty of both—artificial intelligence is here to stay. I’ve found it enormously useful for various projects, especially when it comes to summarizing large amounts of information I’ve written.
For instance, I fed into it my latest book manuscript and asked for 150-word summaries of each chapter. The summaries were 95% accurate, and with maybe thirty minutes of editing they were helpful enough to share with sales and marketing people at my publisher and I’m confident will help others—for instance, people wanting to do a group study of the book.
On the negative side, someone asked me for some additional biographical info, so I asked AI, thinking it would be reasonably accurate by piecing together things online, especially from my blogs and books and our ministry website. But there were a number of inaccuracies in what it said, including that I attended Oregon State University (never have), majoring in business administration (never did). I can’t figure out what sources would have made these claims in the first place, which is one of the mysterious aspects of AI, making us personalize it, saying things like “I get the feeling AI is making this up” or “AI seems to be deliberately misleading me.”
AI has also informed me that I got remarried a few years ago, two years after Nanci died, which was quite surprising, as I think I would have been aware of that. I feel like I need to ask my [nonexistent] new wife for forgiveness since I’ve clearly been neglecting her! Not only have I not been taking her out for dinner, I’ve never met her nor do I know her name! (Though one source did give me her name, which I didn’t recognize.) Generally, I find AI to be 80-90% accurate, but unless I have firsthand knowledge, how do I know which 10-20% is inaccurate?
This is one of the interesting things about AI—sometimes it confidently (again, I am personalizing it) states false information. Why did it not say I had attended Multnomah Bible College and Western Seminary, when there are many online sources that say I did? If it was less than certain (personalizing again), which had to be the case, why not say, “Alcorn may have attended Oregon State and may have majored in business administration,” which would at least serve as a yellow flag, suggesting “Don’t put this in the speaker bio without further confirmation!” Likewise, why did it not say, “It’s possible Alcorn has remarried”? What’s remarkable is, If I hadn’t lived my own life, I would not know what about me was true and untrue!
I directed AI to come up with some questions on chapters I’d written. Some of the questions weren’t good, most were okay, and some were great. So I deleted the okay questions and some of the good ones, replacing them with my own questions which I thought were better. Hence instead of AI editing me, I edited AI. I was the researcher and writer of the book, but this was a time saver and also allowed me to do what I never would have had time to do. This is very different than asking AI to write something I should have researched and written myself!
If AI did the writing for me, I would feel like I needed to credit it just as if I were quoting a source on the web. My recommendation is that if you want to come up with questions for discussion on a book you’re leading a study of, first come up with your own questions, and then ask AI for other questions. I believe that if I am the primary researcher and writer, and if AI is a secondary help, that is the best and natural order of things. But were AI to become the primary and I the secondary, I would no longer be a writer but merely the programmer/instructor of a writing machine. Those, to me, are two radically different things. And they are the difference between true human intelligence that uses a tool, and artificial intelligence that can become a poor substitute for what it means to be created in God’s image.
John Piper gave a helpful answer to how a Christian can wisely approach AI use. He encourages us to ask, “How can I most effectively benefit from its potentials, and avoid its pitfalls?”
Also see John’s answer to “Should I Use AI to Help Me Write Sermons?”
A good rule of thumb I’ve seen mentioned is that we shouldn’t ask AI to do what we reasonably (and ethically) wouldn’t ask other people to do for us. For example, it is reasonable to ask someone to help compile a grocery list, proofread a paper, brainstorm ideas, summarize information, or assist you in balancing your budget if you’re stuck. But it’s cheating for a student to ask a fellow student to write their paper for them, and therefore, they shouldn’t use AI to do that. It’s unethical for an employee to rip off another person’s report and put their own name on it, acting as if it were their own work and effort, so likewise, if you employ AI for projects, you must acknowledge you have done so, crediting it as you would credit a person. If you wouldn’t do that, probably because you don’t want to admit you were lazy, maybe it will push you to do the work yourself and learn valuable life skills!
In her article “3 Questions to Evaluate AI Use,” Bethany Broderick encourages readers to ask:
1. Am I using AI to push past my God-given limits?
“With each new technological advance, we’re tempted to believe that God’s good boundaries no longer apply—that we can know more, do more, and exercise more control than we were meant to. If we’re not careful, we can use AI in an attempt to be gods over our own lives.”
2. Am I using AI to replace wisdom gained from in-person relationships?
“Our AI use should never hinder the real human relationships for which God designed us.”
3. Am I using AI to neglect the work God has entrusted to me?
“AI should never lead us to sloth—allowing technology to do what God has given us the energy and ability to do ourselves. Instead, believers can use AI to support, not supplant, the good work God sets before us.”
Read the whole article here.
