(Photo by Volodymyr Hryshchenko on Unsplash)

I used to say that models are factory machines that output decisions instead of widgets.

I've started to replace "decisions" with "opinions" because every answer a model gives you implicitly starts with "I think that …" or "as far as I can tell…"

Keep that in mind when using LLMs and genAI chatbots. The text they emit can sound firm and confident to your ear, which can give that text undue influence in your company's decisions.

Try to imagine those same words uttered with a shrug, preceded with "I dunno, maybe …", or with a question mark on the end.