A fully automated, on demand, personalized con man, ready to lie to you about any topic you want doesn’t really seem like an ideal product. I don’t think that’s what the developers of these LLMs set out to make when they created them either. However, I’ve seen this behavior to a certain extent in every LLM I’ve interacted with. One of my favorite examples was a particularly small-parameter version of Llama (I believe it was Llama-3.1-8B) confidently insisting to me that Walt Disney invented the Matterhorn (like, the actual mountain) for Disneyland. Now, this is something along the lines of what people have been calling “hallucinations” in LLMs, but the fact that it would not admit that it was wrong when confronted and used confident language to try to convince me that it was right, is what pushes that particular case across the boundary to what I would call “con-behavior”. Assertiveness is not always a property of this behavior, though. Lately, OpenAI (and I’m sure other developers) have been training their LLMs to be more “agreeable” and to acquiesce to the user more often. This doesn’t eliminate this con-behavior, though. I’d like to show you another example of this con-behavior that is much more problematic.