Yes, exactly. With Search, one evaluates a source or two, weighs veracity, reconsiders the current problem given the new information, then comes to a conclusion. All that's out the window with LLMs. Just trust the output without any critical thought. Even worse, just paste the output to a friend or coworker like it's authoritative.

Here's a quick story: I have one colleague who ran some source code through an LLM to summarize what it was doing. Then they copied the output, pasted it into a document, then claim they "wrote" it and wondered where we could store this document for future reference 🗑️

Ew