Uh, Matt, .........So you're saying the "simple but not too simple" isn't exactly just a way to improve AGI prediction, but much more deeper? Like below?:
1) Short program/brain = 2) better prediction: Short prediction = 3) less error: Short error = 4) big paycheck and big happy Amazing. Note that the 1st item above "short program/brain" is really just an item 2 "short prediction", both are short strings/ ideas/ machines. As for the 3rd item, hmm, less error is just shorter program/brain, blending in the brain creates compressed representations, and not saving long strings is part of that blending - it saves on storage space/ time, essentially short programs/brains result in smaller predictions i.e. smaller programs/brains....a loop! So item1 causes a not just actually item2 but a better item1 aka item3. Smaller brain creates smaller brain! Smarter! So, short strings are what you want to store, usually. This IS item 1, 2, and 3. Short, SHORT things are better. Therefore, I'm not sure why Matt "short" is deeper, its not, storing/using short strings improves prediction because good ideas are not infinitely long, and yes, this creates a smaller brain size (more short), but so do other ways to improve prediction accuracy such as not storing similar words or the same word twice (representations). Hmm, all AI mechanisms make the brain smaller/ compress patterns, so they all favor the Short Is Better principal. Not sure if Occam's Razor means that though, maybe it just means the ignore long ideas. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T37756381803ac879-Md5081da017ae04b799faf3f8 Delivery options: https://agi.topicbox.com/groups/agi/subscription
