I bet that, basically, an LLM is just a part of a future AI. The same way that a distributor is part of an internal combustion engine. Or the way Minsk’s descri

I bet that, basically, an LLM is just a part of a future AI. The same way that a distributor is part of an internal combustion engine. Or the way Minsk’s described The Society of Mind. Eager to see if an LLM can be bolted on to some new kind of giant model that does something like mitigate the LLMs weaknesses. Maybe it’ll be a collection of radically different models working together in 20 years, and not just a single model. Like, uh, our own brains. It reminds me of how lex and yacc are super important as grammar and tokenizer, but are only just the front end to much larger projects.