You obviously shouldn’t use LLMs to think, but they are essentially machines that parrot the average Internet user (or in Grok’s case, the average Internet Nazi). Using them to get a tentative guess at what Internet users say isn’t the worst idea.