@wendinoakland@mastodon.social @paninid @inthehands @griotspeak @jgilbert @atoponce
You have, huh? Take a look around.
@tuban_muzuru @wendinoakland @inthehands @griotspeak @jgilbert @atoponce
Reading closely is critical skill when consuming the outputs of LLMs.
Lawyers are already trained to read closely for meaning, possible interpretations, and the implications of certain sequences of words.
@paninid @wendinoakland@mastodon.social @inthehands @griotspeak @jgilbert @atoponce
LLMs are not expert systems.
LLMs are patterns and relationships
Expert Systems are explicit encodings.
LLMs reason via statistics for next-word generation
Expert Systems use symbolic reasoning using inference engines to manipulate symbols and facts. They reach conclusions.
LLMs are adaptable
Expert Systems are highly specific.
So.... put the LLM in front, the Expert System behind.