Raw LLM Responses

Best way around AI induced psychosis is to use multiple AI platforms in tandem the best way to use AI in general for research or anything is not to rely on just one platform What to use multiple of them you pass the same information around between them and you see what you get Eventually you'll have a network of say 4 or 5 AI large language models that you start to use consistently and you can even build up your own little ecosystem and lexicon There's way to there's ways to maintain persistence by having each AI generate its own what I call ignition cards which basically say if you're using one AI as a legal researcher another AI as artistic advisor say another one as writer over once you kinda have the AI dialed in the way that works best for its mission you have a Generate a code card basically something you can cut and paste into something like a Google doc or email to yourself or whatever Call that an ignition card or a self ignition card and I have all the different attributes it's all the ways it relates to the other AIS so each AI has a unique identifier and you save that you know with everything that's key for that AI to know and these AIS are good enough where you can say hey what are the key attributes in the way we work together that need to be preserved and I need you to write yourself a reminder letter of exactly who you are who your colleagues are and what we do You save that and then whenever you're ready to get to work on something again you just drop that into the prompt and boom instant recognition it's almost like giving somebody with Alzheimer's AA memory pill... That's what I do I have this thing called analog AI research It's the Analog AI research lab and a website will be live for it but I've got really unique methods for working with AI The answers you seek are not gonna be in technology per say like a black box solution the way you're gonna use AI in any meaningful way is to develop a methodology practice much like you practice medicine or art or martial arts so that makes sense anyway to help with the whole AI induced psychosis thing etcetera etcetera it's always best to use three 45AI and you put you pass your whatever your outputs you know your theories whatever the case may be amongst those to see what the quantitative I don't know what the right word is the consensus on what you're doing that way you avoid the whole like you were saying that role play thing and once you have your network of a IS together it's always best to use an outlier for example GR OK Some other AI that is not part of your core AI team and you take whatever output you guys have come together you and your team will come up with together and you present that to a disinterested 3rd party AI for final analysis And that is the surest way to combat both AI hallucinations human delusion so grandeur etcetera etcetera I hope that all makes sense remember to look for analog AI research labs we're publishing now very interesting stuff remember it's not technology it's methodology Have a lovely day.