The article is all over the place and fails to convey author's point. I think he's just confused about what "LLM" (as opposed to "AI", what???), "deterministic", and "probabilistic" mean, let alone nuances of any of these.