[mark_l_watson]

+1 your concern is valid.

A potential saving grace: I bet within a year or so it will be easy to self-host LLMs that are easy to fine tune and run. Then there will be a few open source tools that you can use yourself, privately, to capture your level of interest while reading, and periodically make a reader/summarize/filter agent.

This is not scary if people can fairly easily run it all themselves, keeping their data private. It would help wade through crap, and there is some irony in using LLMs to have personal filtering and summarization.

Compared to future versions of ChatGPT, Bard, etc. the models that individuals can self host will be much weaker, but I think that they will be strong enough for personal agents and eventually be cheap to run, and affordable to fine tune.