Increasingly we turn to a familiar AI platform such as ChatGPT, Gemini, or Claude for all kinds of things. Yet you don’t know whether you can really trust the info that it sends you.
Early on when ChatGPT first appeared, I asked it about myself. It quickly gave fairly accurate biographical information, though I was surprised to find out that my PhD was from the University of Minnesota, a campus I have visited only once in my life. How nice to get a PhD from that brief encounter! But when I put in my wife’s name, I was saddened to see that she had died in 2019. When I informed her about her death, she was surprised, since no one had told her. I promised to split the insurance money.
In a book that I just finished writing I didn’t use AI at all. It is a book for Oxford on Religion and Violence in a series where they ask people they regard as experts in a field to give a quick primer on the basics for a general public. I had no need for AI since I knew most of this stuff and had researched it myself, usually field research where you actually go somewhere and talk with people, and not just read what other people have written about it. So I’m a big fan of expertise. I’m appalled at the decline of trust in expertise in American society, especially in medical matters but also economic ones, security issues, intelligence gathering, and the like.
AI replaces trust in expertise, though paradoxically it totally relies on it. AI info is an amalgam of purloined information from everywhere, including, I’ve been told, several of my own books. So what you’re getting with AI is a distillation of expertise, though without verification or cross-checking. That’s where the user has to take control.
Like any research source, before you claim something is true you have to verify the info you get from AI. I do this in two ways. One is to use several AI platforms, such as the common ChatGPT, Gemini and Claude, and have them critique one another. The other is to go back to AI’s sources (it will often give a citation link, or you may need to ask them where they got that info) and see whether it is reliable.
I applied this litmus test to the information I recently received from AI. I was writing a piece on my religious evolution and showing how it mirrors in many ways the changes in religiosity in American society, particularly Protestant Christianity, from the Eisenhower years of the 50s to the Trump years of our times. I wanted to get more information about my ancestors German roots. I had heard from my mother that our German ancestors had been part of a pietist group in Germany in the Lippe region near Munster where most of the mid-19th century German immigrants to Warren County, Missouri originated.
I turned to AI to see if it could give any clarification on the German Juergensmeyers’ religion. It seems that a group of young Germans in the farming region of Lippe just northeast of the city of Münster was disaffected for several reasons. One was economic—the population was growing and new plots of land were difficult to obtain—and another was political. The revolution of 1848 that promised political freedom and economic opportunities was stifled. And accompanying that new repression was a religious aspect: a new wave of pietist Protestantism, the evangelisch, “Evangelical,” movement. In the mid-19th century it was influenced by erweckungsbewegung, a movement of religious revival or awakening that emphasized personal piety and social service. It was somewhat out of line with the more conservative Reformed Protestant churches that were dominant in the region, and there were some restrictions on where and how the Evangelicals could worship
This meant that the evangelical faith of the immigrants was in line with American Methodist piety. Moreover, the Methodist church had sent a horde of German-speaking preachers sent specifically to convert German immigrants in places such as Warren County in Missouri where my ancestors settled in 1860. This explained why our ancestors so quickly became Methodists.
As with all AI information I cross-checked and tried to verify. In one case I found that the AI platform got its info from family memorabilia, not very reliable. But in another case it was a report for the state of Wisconsin on German immigrants specifically from the Lippe region near Münster from which my ancestors came. This was a more reliable source, but one that still needed cross-checking.
So AI has its place. It can be maddening, however, when you’re trying to teach, because you don’t know where students got the information for the term papers or whether they even wrote them themselves. In my classes at Claremont where I’ve taught one course a week after retiring from UC, all of my written assignments (tests, term papers) were in-class written with pen and paper, or in the form of oral presentations. Nothing digital allowed. AI has changed our teaching style, probably for the better, I think. Yet in my class I’ve had sessions on how to use AI responsibly, since it’s here to stay and is a valuable resource for many things. Since people are going to use it, they need to use it responsibly.
I said I didn’t use AI for the book I just finished, though that’s not quite true. I used it to help find citations. I remembered the theologian and my former professor, Reinhold Niebuhr, once saying that if violence is to be used for military or police protection of the public order, it should be as swift and sure as a surgeon’s knife. I had forgotten in what book, much less what page number that quote was on, but AI found it in a nano-second. And since I had Niebuhr’s books on hand I could quickly confirm it.
The upshot is that AI is no better or worse than any other source of information we get. It may be faster, but not necessarily more reliable. Expertise is a wonderful thing, even when it is churned through the algorithms of AI. But we still have some work to do. The word verify comes from veritus, truth. It’s an elusive concept but one worthy of trying to achieve.