There is a qualitative difference to the argument: lectures (from before millennia), books (in the last few hundred years), and movies (in the last few decades), were all non-interactive. A student would sit and consume passively. There is a limit to how well you can learn with passive instruction alone. The promise of AI is to bring the Oxford model - 1:1 studying with a tutor, with whom the student engages and to whom the tutor gives their full attention - down to a price point at which it would become feasible to educate the masses with it.
Are LLMs actually capable of re-producing a good-enough simulation of the Oxford model? Will this simulation of the Oxford model produce better educational outcomes at scale - something that has been theorized but was of course far too expensive to prove - when deployed to the masses through public education? Maybe? Who knows? Time will tell.
For what it's worth - I don't think books will become obsolete, either in schools or anywhere else. Paper books might become obsolete, but there is no option on the table that proposes to completely replace the art of science of having life experience and committing it to longform. As long as copyright protects its commercialibility, it will continue.
Are LLMs actually capable of re-producing a good-enough simulation of the Oxford model? Will this simulation of the Oxford model produce better educational outcomes at scale - something that has been theorized but was of course far too expensive to prove - when deployed to the masses through public education? Maybe? Who knows? Time will tell.
For what it's worth - I don't think books will become obsolete, either in schools or anywhere else. Paper books might become obsolete, but there is no option on the table that proposes to completely replace the art of science of having life experience and committing it to longform. As long as copyright protects its commercialibility, it will continue.