The other issue is the one you've raised as have a few others -- LLMs can be very useful for experts who can catch errors and refine responses, but bootstrapping conceptual expertise when novices start with LLMs on day 1 is going to be a hard pedagogical challenge.

We were discussing this issue at the @cu-esiil.bsky.social meeting a few weeks ago. On one hand, LLMs seem like just higher level abstractions, like going from assembly to Fortran to R. On the other, they are probabilistic not deterministic, which makes them different.

So if Sanger capillary is the Greatest Generation (sequenced human genome primer walking uphill both ways thankyouverymuch), and AVITI is Gen alpha, does that make 454 Gen x or the millennials