undefined | Better HN

0 pointsGardenLetter277mo ago0 comments

And isn't the obvious solution to just make some sort of browsers add-on for the LLM summary so the request comes from your browser and then gets sent to the LLM?

I think the main concern here is the huge amount of traffic from crawling just for content for pre-training.

0 comments

otterley7mo ago

Why would a personal browser have to crawl fewer pages than the agent’s mechanism? If anything, the agent would be more efficient because it could cache the content for others to use. In the situation we’re talking about, the AI engine is behaving essentially like a caching proxy—just like a CDN.

j / k navigate · click thread line to collapse

0 pointsGardenLetter277mo ago0 comments

And isn't the obvious solution to just make some sort of browsers add-on for the LLM summary so the request comes from your browser and then gets sent to the LLM?

I think the main concern here is the huge amount of traffic from crawling just for content for pre-training.

0 comments

otterley7mo ago

j / k navigate · click thread line to collapse