Don’t bother asking Google if a product is worth it; it will likely recommend buying whatever you show interest in—even if the product doesn’t exist.

This seems like a general problem with these LLMs. Sometimes when I'm programming I ask the AI what it thinks about how I propose to approach some design issue or problem. It pretty much always encourages me to do what I proposed to do, and tells me it's a good approach. So I'm using it less and less because it seems the LLMs are encouraged to agree with the user and sound positive all the time. I'm fairly sure my ideas aren't always good. In the end I'll be discovering the pitfalls for myself with or without time wasted asking the LLM.

The same thing seems to happen when people try to use an LLM as a therapist. The LLM is overly encouraging and agreeable, and it sends people down deep rabbit holes of delusion.