AI and the Catholic Moment, by R.J. Snell

By R.J. Snell, The Catholic Thing, July 14, 2025

R. J. Snell is editor-in-chief of Public Discourse and director of academic programs at the Witherspoon Institute in Princeton, NJ.

Elon Musk recently promised to improve his version of an AI chatbot, nicknamed Grok. The fix did not go well, to put it mildly. In one description, Grok was intentionally designed to be “an ‘anti-woke’ competitor to OpenAI’s ChatGPT.” But Grok seemed indistinguishable from its competitors. The fix, as was widely noted in recent days, whatever else it was, certainly was “anti-woke,” with Grok turning into a holocaust denier, casting aspersions against Jews, threatening the Turkish president, and Grok giving itself the new nickname of “MechaHitler.”

More than anything, the episode reveals, as sober-minded people know, that artificial intelligence is not intelligent. Large language models scour pre-existing texts and are highly efficient at pattern recognition. But they entirely lack the capacity for judgment or an experience of reality allowing them to engage in those second-order reflections typical of rational entities to see if a statement corresponds in any meaningful sense with primary reality as reality presents itself for our understanding and judgment. The so-called “hallucinations” of AI are good indicators of this fact. …