@coriander decades if not centuries of research into furmal models of language have only ever yielded more questions than answers
@aescling The best we've come to natural language through computers is LLMs doing predictive algorithms, which are absolutely nothing like actual language in any way
@coriander i somewhat contest this because it depends on how you define “best”; LLMs are not remotely informed by Real Linguistic Theory and you can go pretty interesting places translating that shit into something computers understand. (machine learning can even be done in ways that is at least infurmed by ideas in the field; i’ve done contact work annotating sentences fur some kind of speech recognition system that never ended up going anywhere…) which honestly just makes LLMs piss me off way more; it’s just the completely wrong direction fur this kind of research and application imo
@coriander i’ll fully admit part of this is like, pettiness, but LLMs feel like the industry smashing a giant hellish probabalistic sledgehammer to Ideas and Theory and Math that excite me. and it makes me so sad
@aescling @coriander this guy is also AI's number one hater which is cool