18
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 21 May 2026
18 points (100.0% liked)
Just Post
1500 readers
63 users here now
Just post something 💛
Lemmy's general purpose discussion community with no specific topic.
Sitewide lemmy.world rules apply here.
Additionally, this is a no AI content community. We are here for human interaction, not AI slop! Posts or comments flagged as AI generated will be removed.
founded 2 years ago
MODERATORS
human knowledge, not intelligence. yes it involves absolutely massive quantities of sloppified pigshit but it also includes practically the entire corpus of published literature, both academic and entertainment.
either way my point is that you can't bruteforce an artificial brain, but these trillion dollar corporations are still all-in on 'scaling will solve all our problems'.
They actually don’t even have all that. LLMs run into the same issue that human beings do when companies are scraping web content: paywalls.
There have been some articles about it recently. But basically they have an established heavy bias towards content that is freely available, which tends to be lower quality reporting, thinly veiled advertisements, or outright propaganda. The idea that LLMs are at least exposed to the plurality of quality knowledge is itself just a marketing push, in reality there are glaring gaps
The fun part is comes with future training via scraping the internet. At this point, the vast majority of new articles are LLM-infused at minimum. I'm sure smart AI scientists are trying to curate the training data, but I have no idea how they're avoiding incestuously training LLMs on themselves. Otherwise the future it seems like we'll get is a bland, grey Hapsburg training set (read: internet) and even worse LLMs
all good points.
i'm just gonna repeat myself though that the solution they're selling is scaling, which means ultimately the goal is to absorb not just as much information as possible, but all the information, which is what i was referencing in my first comment.
Gotcha. Yeah I mean I guess even given it really does have access to everything at some point, I feel like we users would just end up in a Library of Babel situation.
Which we somewhat already are in with the modern internet, yet at this point in time it is still possible to exercise discernment and verify the integrity of sources. If we end up restricted to “conversations” with chatbots our connection to truth will decay rapidly. Probably why they’re pushing it so hard tbh