Post

Nature published another pile of trash i am trying to catch up on some of my reading but this one is getting under my skin so here’s a thread highlighting why this piece is either ill-informed or intentionally ignorant of a wealth of knowledge from embodied cog sci and related fields 1/

It start by misinterpreting Turing. the Turing test only assess if a human can be fooled by a machine (a question that has been rendered meaningless, imo) (claim from the paper and the first page of Turing’s 1950 paper side by side) 2/

behold... "llms can produce human like text, thus llms have human-level intelligence" (disclaimer: don’t expect anything but snarky comments from me today) 3/

a chatbot can … checks note… chat 24/7 with hundreds of millions of people around the world. mind blown (I address shortcomings of the other arguments later) 4/

gee, i wonder why 76% of experts don’t buy into a scientifically bogus idea with nothing but multibillion dollar marketing hype powering it 5/

well, that settles decades of investigation into what intelligence is! all those cognitive scientists, psychologists, neuroscientists, anthropologists can pack and go home now 6/

alright, they gonna clarify everything, right (spoiler, set yourself up for disappointment) 7/

*insert the goose chase meme* who thinks this… who, who? 8/

neat conclusion! too bad they have so far provided no convincing evidence to support it 10/

this is a strawman argument: "Einstein is the gold-standard of perfection but (physicists) aren’t perfect like Einstein and thus neither should AGI be" 11/

"a single individual can’t do everything thus AI systems trained on trillions of tokens harvested and aggregated from billions of human interaction data can’t be expected to do everything"