'Whatever #LLMs model, it is not human language, which is not a "large and growing heap, but more a flowing river."'
There is "the assumption of language completeness – there exists a 'thing' called a 'language' that is complete, stable, quantifiable, and available for extraction from traces in the environment.
And there is "the assumption of data completeness – all the essential characteristics can be represented in the datasets that are used to 'train' the model."
https://www.theregister.com/2024/08/30/ai_language_cognition_research
=> More informations about this toot | More toots from johnpettigrew@wandering.shop
text/gemini
This content has been proxied by September (3851b).