When we say that
>we are asking the model to remind us of the lyrics of a well-known nursery rhyme, ... what we are really doing is asking it the following question:
>>Given the statistical distribution of words in the public corpus, what words are most likely to follow the sequence “Twinkle twinkle ”?
To which an accurate answer is “little star” 😀

arxiv.org/abs/2212.03551

Follow

is a necessary condition and comes before which is in turn necessary for the acquisition of :

>Language use is an aspect of human collective behaviour, and it only makes sense in the wider context of the human social activity of which it forms a part (*Wittgenstein, 1953*). A human infant is born into a community of language users with which it shares a world, and it acquires language by interacting with this community and with the world it shares with them.

Trying to "reverse engineer" intelligence from LLMs seems like a futile endeavor. Instead of , it may be more useful to redirect efforts to building ()

Sign in to participate in the conversation
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.