I haven't read the article, but it sounds to me you're conflating “how much do regular users trust LLMs to produce good/correct output” with “how much do capitalists trust LLMs to become (and remain) profitable”.