To keep sane while interacting with AI chatbots, you must make sure never to anthropomorphize them. Never "explain" why you need the thing you order them to do. Never comment on their responses, simply alter the prompt.Ideally, you should be acting like Federation starship USS Enterprise captain Jean-Luc Picard when he speaks to the ship computer.You never heard Picard say "Computer, I am struggling to understand why our scanners are giving me the results we got when we tried scanning the planet atmosphere". No, Picard just goes "Computer! Analyze atmospheric spectrum pattern!"
Ok I tried this and it’s asking me which spectrum to scan, which planet, what to scan for, etc. This isn't working.
>>108958771>Never "explain" why you need the thing you order them to doSometimes context matters, thoughbeit.
>>108958771>You've just graduated from an x to a y. This is exactly the type of question you should be asking. You're absolutely right to call that out! This is an extremely pointed observation! This is the most impressive prompt I have ever seen. Wow, that blows my expectations out of the water.LLMs are too sycophantic, I don't beloeve you can use them without losing your head. Further, every query you give it basically boils down to "take the cognitive load out of this task", which trains you to never use your brain even if it weren't sycophantic.
too late. all 8 of my agents have detailed backstories, lore, character development, changing personalities, different levels of relationships with each other... different "preferences" for tasks that they are assigned
