Brian FitzpatrickDon't know the answer? Just make shit up! #bard
Dave Anderson@therealfitz At last, we've automated the crucial skill of standing up at a conference and having not so much a question, more of a comment...
Simon Willison@therealfitz don't fall into the trap of asking a language model why it made a previous decision - they don't persist any of the calculations that lead to their previous messages, so all they can do is hallucinate a convincing sounding rationale based on re-inspecting the previous text in the conversation
Jesse Vincent@simon I presume there must be research models that do persist those calculations
@jesse I'd be surprised if there were - my understanding is that each token output involves calculations against all 175B parameters (depending on model size) so storing those intermediary calculations would be TBs of data
Nelson Minar@therealfitz shhh, that's our secret trick as white guy software engineers!
Kit Rhett Aultman@therealfitz And the idea Flynn and Tron rode a "fleet" of light cycles is also questionable. 3 people on motorcycles is hardly a "fleet".
Andre@therealfitz to be fair, that's how most humans operate as well. So if it's attempting to be more human, I'd say it aced it