@YiMaTweets I can see that's the intended road that was being walked, but is there a strong argument to be made that the two roads do not eventually merge and end at the same destination? GPT3 appears to have symbolic understanding and ability to do some reasoning.
@YiMaTweets Words are references to conceptual understanding. It's true that not all memories are addressed yet with a word reference, but it is hypothetically possible to do so in a machine intelligence? Language is infinitely expandable (barring technical limitations).
@Jake_Browning00 Ideally a model will also capture the relationships between words and other sensations. In fact, how better to express those relationships than an expanded language? (2/2)
@Jake_Browning00 This is a nice write up, although I'm not sure I agree with the assertion in total. Language is certainly a limiting factor to understanding, assuming that language is incapable of expansion. LLMs capture the semantics of language through relationships between words. (1/2)
@mark_riedl If you believe in the technology and have something unique to bring to the table why wouldn't you? Chase your dreams, have an ounce of passion, commitment, and confidence. Many engineers lack confidence in themselves.