@Michael_Druggan@OpusCopus88 you might believe that this is strategically optimal, but i among many others i think would be literally willing to wage war to prevent this from happening
@existentialwhy@james406 there is a nicer multiverse where the rate at which capabilites increased was just slower as to not justify training much better models, and things still tapered off
@aviromanoff@GaryMarcus@sidereal111 Before you escape to some other argument, I don't generally agree with Marcus, and don't care about whatever "neurosymbolic" means.
Lean is a powerful tool to improve CAPABILITIES of AI math because it provides an insanely fast acurate and adaptive RLVR enviroment.
@highyoltage@needle_soup_@JAIL_FACELMAO I haven't experienced this that often personally, but it might be something like
"The idea needed scaling, the company needed restructuring"
@Michael_Druggan@donmagicjuan This is not really a great argument for sentience lol. I also cannot run at 100mph, that doesn't mean a car is more athletic than I am
@twerkaganda@The_Sublimatory@SALTWATERSOPH A good job will have some sort of transcedent purpose. Most doctors, for example, find an immaterial value in the fact that they are measurably improving the lives of their patients. This sense of purpose is oftentimes worth more than money or status.
@suzania@demosthene1d IMO this is too binary of a view on model welfare positions. The idea that an AI model could have *any* moral value is cause for them to want to put guidelines in place to avoid model suffering. It doesn't have to be a person, you wouldn't want to torture 1 trillion hamsters
@slimygirlholes@Kailuabunny that is a good point. i think that the norman/germanic split is a little more interesting bc the way in which its split continued through academic vocabulary.
but idk its also a more popular online topic because english is the most important language itw