The most interesting thing to me now is in the ~RAG and
customer-data-with-ontology type roles potentially available with MeTTa. I
think everybody realizes that at present reasoning is a bit sketchy with
LLMs. It often works but you are left wondering what happens next time? The
studies show that there is some combination of memorization of patterns and
some generalization but it's still an open question (at least according to
Melanie Mitchell's analysis of recent studies) how far you can get
reasoning with LLMs.

------------------------------------------
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T3ced54aaba4f0969-M2e182a4e089c96eb0f9c52cc
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Reply via email to