Ad-homineLLM: Are you wrong because you used an LLM?

It’s unfortunate that the term Ad hominem translates directly from Latin to “to the person” in English, since the heart of the fallacy concerns the irrelevance of the source, rather than the humanity of the speaker. The general concept is attacking the source of an argument versus addressing the argument itself. Aristotle could never have envisioned something like an LLM, but he was surrounded by people who believed in all kinds of deities.

There are plenty of valid reasons to dislike the output of an LLM. The output can be inaccurate, generic, or just plain verbose. Even some wishy-washy “I don’t like the tone” is a better argument than being dismissive because an LLM was involved in the creation.

It’s also important to recognize the asymmetry of being tasked to validate arguments that took very little time or effort to craft. Perhaps an argument is too bloated and you’re not willing to read pages of verbose text? Make that argument instead of resorting to tribalism about the type of skin the presenter lacks. Many humans have fought to the death because of varying skin colors: Do the rules of logic change when NULL is now in the mix?

Note: Credit to HN user api for the term “Ad hominllm” and to both egberts1 and tmaly for adapting the term “Ad machinam”, each within 10 minutes of one another!