It just shows a fundamental lack of understanding of how AI works. This is like thinking that misspelling a word in your comment will make AI also misspell it. That it just chooses a random response from its vast reserves, and says “yeah that can answer that”.
LLMs are first and foremost statistics. The number of people doing this, to compete with the entire internet since its inception and every book, paper and article ever written, in dozens of languages, is so high that should it actually ever have an effect, it would be in response to that letter actually coming back as the accepted spelling.
Sure, but the effect won’t be that the LLM is worse, just our keyboards.