I was talking to my AI on my drive in to work today about how *meaning itself* is almost a dynamical evolving thing. If we take a frontier model trained today, and then take it 50 years into the future, it won’t simply be factually incorrect on a lot of things and completely outdated in others… it will be semantically inaccurate as well. The geometry and interrelationship of meaning will have evolved to a point where it no longer captures the contemporary status of understanding. Relationship between words will not be the same even if the definitions are.
I’m at work just skimming but I think that I share your view on the meaning that we compress to words. I love this and will respond after I read it. Thanks Kevin.
Perhaps now drop in my work as a nonlinear dynamical system. And simply ask: how does this fit in? :)
https://kevinhaylett.substack.com/p/geofinitism-language-as-a-nonlinear?r=6479k6&utm_campaign=post-expanded-share&utm_medium=web
I was talking to my AI on my drive in to work today about how *meaning itself* is almost a dynamical evolving thing. If we take a frontier model trained today, and then take it 50 years into the future, it won’t simply be factually incorrect on a lot of things and completely outdated in others… it will be semantically inaccurate as well. The geometry and interrelationship of meaning will have evolved to a point where it no longer captures the contemporary status of understanding. Relationship between words will not be the same even if the definitions are.
I’m at work just skimming but I think that I share your view on the meaning that we compress to words. I love this and will respond after I read it. Thanks Kevin.