You can’t imitation-learn how to continual-learn (opens in new tab)

<p><span>In this post, I’m trying to put forward a narrow, pedagogical point, one that comes up mainly when I’m arguing in favor of LLMs having limitations that human learning does not. (E.g. </span><a href="https://www.lesswrong.com/posts/ZJZZEuPFKeEdkrRyf/why-we-should-expect-ruthless-sociopath-asi?commentId=5js8FmkJPQMoQ6SWv" rel="noreferrer"><span>here</span></a><span>, </span><a href="https://x.com/steve47285/status/2031423263558054155"><span>here</span></a><span>, </span><a href="https:...