All good writing is born of frustration. The writer stumbles upon a truth that needs to be said, which forces him out of complacency and into work. For example, why do people not change their minds when faced with unexpected evidence? For me, originally trained as a scientist, that’s the greatest frustration of all.

This personal stake makes me wonder why anybody would let AI write the first draft. It would only flatten out my idea by giving me the most likely response. This is not to say that I don’t use AI. Indeed, I use it a lot. Sometimes, I give it my text, one paragraph at a time, and ask for comments. They can be profoundly challenging and force me to improve my writing. But AI impresses me the most when I use it for research. It finds all my sloppy simplifications and gross generalizations and forces me to deal with what I would know if I were an expert in everything.

With AI, the accumulated knowledge has been given a voice that we can invite to present us with the middle-of-the-road answers. Its output includes any bias and errors already present in the information fed into it. But unlike AI, human thought does more than recapitulate known facts. It asks questions, evaluates answers, notices falsehood, and, most importantly, seeks truth.

It is an article of Christian faith that all truth is in harmony, and when it is reached, then questioning can rest. Deception may have local harmony, such as in conspiracy theories that carefully select which evidence to admit and which to exclude. But faith tells us that the true answer will be fitting with every other true answer. Global harmony, even across disciplinary boundaries, rather than the exclusion of what does not fit, is what we must look for.

Since AI is trained with a widely disparate body of knowledge, interdisciplinarity is deeply embedded in it. While it carries over the errors and biases of its training data, these are not likely to be consistent across disciplinary boundaries, and a careful reader of the AI output can notice this. When we probe these inconsistencies with follow-up questions, AI becomes a powerful tool to avoid deception and gain true understanding.

The human power to question until inconsistencies are resolved lets truth find you and transform you. It is an act of faith to accept the resulting responsibility. Therefore, when the evidence proves you wrong, you should change your mind, even if we find this very hard. In the working of the human mind, established beliefs tend to carry more weight than newly acquired information. It might be the reason why it is said that scientists, who must question established beliefs when confronted with contrary evidence, do their best work when they are young.

Philosophers, however, might discover that they do their best work when they are old. Indeed, there is a great deal in philosophy that is the wisdom of the ages that we appreciate only late in life. However, what changes is the context of what else we know. For this, we now have the help of AI to help us change our minds when we are applying timeless insights in outdated ways.