Possibly the LLM vendors could bias the models more toward nonprofessional content, but then the quality and utility of the output would suffer. Skip the scientific articles and books, focus on rando internet comments, and you’ll end up with a lot more crap than you already get.