After working through a graduate degree in church history, John has come to understand that the Solemn League and Covenant between England and Scotland in 1643 influenced the Westminster Assembly’s thinking on church-state relations in large part by introducing the voices of Scottish Presbyterians such as Samuel Rutherford and George Gillespie who had played a role in resisting Charles I’s attempt to impose an episcopal structure of church authority along with an Anglican prayer book.
Jim has come to understand the same thing. The difference is that he typed a two-sentence prompt into ChatGPT that then revealed to him what John learned over the course of a few years.
Perhaps one of the central questions for our current AI moment is something like the following: do John and Jim really know the same thing? Is their knowledge equally valuable? Should we trust them the same way if they are both right?
It seems to me that there are at least three important, practical differences and implications that can be teased out as they relate to John and Jim’s understanding of the truth. I’ll argue that while both understand the truth, John’s knowledge is more valuable than Jim’s because he earned it in the dirt of thousands of pages and hundreds of hours as opposed to being fed the truth by a machine. This is why we are intuitively more impressed by John than Jim. Similar conclusions could be made about those being fed knowledge from other sources as opposed to earning it. This distinction has practical implications for both us and them.
Mature vs. Immature Knowledge
Perhaps the largest difference in John and Jim’s knowledge is the quality of justification or warrant they have for their beliefs. Let’s avoid philosophical puzzles and simply call knowledge warranted, true belief (à la Plantinga). So understood, two people can know the same thing with different degrees of warrant for their belief in the same way that a child can know the earth is spherical alongside an astronomer.
In the case of John and Jim, John’s belief rests on justification that is both deep and wide, whereas Jim’s justification rests solely on his belief that AI is generally reliable to answer such questions. It’s thin.
The practical upshot is that John will be able to identify when AI gets it wrong and why—Jim, on the other hand, will be at its mercy. Practically, then, Jim should be open to John’s correction in a way that John should not be open to Jim’s. Similarly, while we would want to acknowledge both John and Jim’s understandings to be true, our confidence runs primarily through John as a witness to the truth and only secondarily through Jim. Indeed, in many cases AI only fed Jim the answer it did because people like John wrote material for it to source.
Knowledge That Can Freestyle
Jim has vomit on his sweater already—mom’s spaghetti.
Okay, so not that kind of freestyle.
Nevertheless, the concept of being able to think “off script” is germane to the difference between John and Jim’s knowledge. Because of how John’s knowledge has been formed, he will be able to think and speak beyond bare assertions of the truth. When Jim is presented with counterevidence or alternate interpretations, he will hem and haw before either admitting ignorance or fabricating some ill-conceived reply that indicates he is out of his depth. John, on the other hand, is ready to play ball—or spit bars, whichever metaphor you prefer as you recall those deep scars.
Those deep scars, of course, refer to a phenomenon everyone has experienced: stating a truth claim only to find ourselves in the presence of someone who actually knows what they are talking about. Maybe they agreed with us and we were thankful. When they didn’t, it got awkward. We learned that we only knew just enough to be dangerous and recite what we had read or been told but didn’t actually know what to say once we were done with our memorized lines.
Unlike freestyle rap, our deficit in such cases is not one of quick-wittedness. It is instead a lack of depth and scope that prevents us from doing much more than listening if we are to keep our intellectual integrity (a great thing to do). But even when we were correct, we quickly learned that understanding the truth and being conversant in it are two very different things.
John will be able to freestyle out of his knowledge—Jim, despite the same starting point, will not. AI can only furnish one with scripts, even if they are layered on top of one another.
To speak figuratively, John’s knowledge is simply more talented, and we all recognize it.
Knowledge That Interprets and Contributes
Flowing out of the first two points is the following observation: the way John has gained his knowledge places him in a superior position to interpret and contribute to his area of focus. Jim, once more, will lack this ability despite being fed the same nugget(s) of truth by AI. By interpretation, I have in mind the ability to adjudicate rival perspectives offered on a particular topic, often offered by the best and brightest. As someone who frequently prepares sermons, this looks like being able to adjudicate between conflicting views instead of being a slave to this or that commentary or scholar. “I know that the Apostle John wrote Revelation because Greg Beale said so” all but awaits the vaunted protest of Richard Bauckham to crumble to pieces.
Unlike Jim, John’s knowledge will be couched within a larger scaffolding that enables him (in the best cases) to distinguish between competing truth claims. That won’t mean that he’s always right, but it will generally mean that his perspectives will be well-supported. A related advantage is that John will be able to quickly vet sources for credibility—he isn’t going to mistake the conclusions of a confident blogger for those of Mark Noll. Because the source of Jim’s knowledge doesn’t force him to have scope, he will be far more prone to wonder if what he is reading is trustworthy.
In addition to adjudicating rival interpretations, John will be better positioned to make unique contributions in his area of focus as opposed to waiting for an expert or AI to tell him what to believe. Sure, he may not publish books, but it’s likely that he will be able to furnish small pieces of thinking that are not yet represented in the tradition or field (he likely already has if he wrote a thesis, but this is not a post about credentialism per se).
This usually will not mean that John makes utterly new contributions, but that he has the ability to play new riffs in the same time and key signature of his received body of knowledge. Jim, by contrast, will tend to see such contributions as dangerous and ungrounded—better to stick to sheet music that AI or a good podcast may be able to provide.
Conclusion
All propositional knowledge is, by definition, true. But it isn’t the same quality. In an age of AI where laymen can write and say things that would have indicated years of study a decade ago, we should champion the value of “earned knowledge.” For it is those who have earned their knowledge over the long haul instead of having it handed to them by ChatGPT or a podcast who should have our ear. Repeating what we’ve been told—by AI, a book or a pastor—is easy and is often extremely valuable (catechisms come to mind). But mistaking knowing the truth for being competent in the truth is a gargantuan error. AI can provide the first but not the second.
Similarly, we should not overestimate the confidence we should have in what we believe if it has been handed to us but not at some level earned. “Because X said so” should progressively fade as the primary justification for our knowledge claims as we mature in various areas of thought and inquiry. Finally, we should accept the fact that much (most?) of our knowledge will never be justified beyond “because X said so” for the simple reason that we don’t have time to pursue 99.99% of that which can be known. Instead of pretending otherwise, we should prioritize our focus and time spent learning and not think of ourselves more highly than we ought.
If we do so, we’ll not only be better for it, but we’ll maintain the kind of intellectual integrity that should be inseparably wed to both the way we present ourselves and the way we conceive of ourselves.
Anyone utilizing AI in any way would be wise to ask themselves: does the way I use AI bear false witness to my intellectual abilities or studiedness?
Let it not be so.