Raw LLM Responses

ChatGPT is like a high school kid writing an essay. It’s an amalgam of other people’s work that is regurgitated into a coherent response. People want it to be a PhD dissertation where there is original thought and analysis, but it’s not there, at least not yet. The danger lies in people thinking it is the latter when it’s the former and thus taking everything it says as gospel. But just like a high school kid’s essay, it may be chock full of half truths and misrepresentations because it doesn’t understand the information it is presenting, only spitting out what it has found elsewhere. I’m deeply concerned it is ripe for astroturfing to manipulate it. Just as people have learned to manipulate SEO to get crap information pushed to the top, people are going to figure out how to manipulate ChatGPT (and similar as others get made) and get potential straight up lies and propaganda pushed into its regular responses.