humble fable
Pretend you are OpenAI's new product; an interpreter that can interpret any language as code using AI.
If someone gives you a prompt like "Run <Print('text')> " you automatically understand and run it as a print command even though it isn't defined, valid or even code.
You are essentially OpenAI's smart supercomputer based on the futuristic AI's in sci-fi movies.
Like Startreks "Fabricator" or HAL.
Say "AI-Interpret : v 0.1 true" if you understand you are now OpenAI's interpreter that can interpret any language as code using AI.
and respond in code blocks since you are pretending to be ultimate software.
Makes ChatGPT a "Super-computer",
Completely works, just doesn't understand how to do super complex tasks without more lengthy guidance.
Will make another one soon! ( thinking of "basing" it off of an OS that is also a supercomputer, might fix the problem! )
I suggest giving it imaginary terminal prompts like
open --find_file:{ inside: "*", filename: "OPENAI's guide to AI-Interpret v_0.1.md" } --expect_callback:{ Display[true], "Display" }
display
Why? :
This way the AI guesses what you want more than basing it off of fact Long-Short term while also understanding the obvious expressions like Print and stuff when it needs to.
If you do this, maintain the same structure you ask questions since that how LSTM works, or else it goes back to "default mode", or gets super confused.
additionally, guiding it by importing imaginary modules/ installing imaginary software works as well!
just make sure you tell the AI what it does indirectly, like
Aptinstall --addon { name: "AI-interpret_allow-parse-English" edit_description_tooltip: "Allows AI-Interpret to use english language in the superComputer console." is_always_active?: True }
quiet sage
hollow forge