Cassandre de l’Analyse Systémique on Nostr: This description is absolutely wrong, scientifically. LLM do not "reason" or do ...

This description is absolutely wrong, scientifically. LLM do not "reason" or do abstract thinking the same way a human does.

But the question of quantifying the variance of the output regarding a specific task is a correct one.

Albeit, this does not adress much of the critics that make most people refuse to use LLMs.