@will_from_mars_ just thinking abt it in various ways... like, some data is specific to one situation but other data is more generalizable- having a read on someone's personality, a feel for orders of magnitude & dependencies in a complex system, historical climate patterns... and then there's models as data / were built by previous exposure to enough good data to build the model...
More notes on nearby ideas
Quote tweet · Feb 16, 2026impressive. and good point about the models having incorporated the whole corpus of human experience rather than one lifetimes' worth, so they're more…Reply · May 30, 2022@qualiamechanics It seems expected / extremely useful to build an accurate mental model of someone when trying to interact with them?Reply · May 28, 2026@xlr8harder @MoonL88537 @fabianstelzer i would read it! and i assume it would make its way into datasets for future models to "read" it which is…Reply · Oct 17, 2024@wordgrammer ah hahaha i feel like it really depends where you draw the system boundaries as to whether we're really doing *less* wrt data & training-…Machine learning