Narrow AND incorrect

Hey, incorrect only the majority of times!

/s

More correct than whatever slop one liners you get from dudebros on the internet

Great, another reason to not like AI…

Ask the teachers about the kids nowadays.

Not since AI, it’s been since internet. No since national school. No, since religious teaching (this one is very dangerous). No since…

You know what, screw this. Knowledge where never the issue, cause the easier the acces is, the less you make it, and if you never process what you understand, you end up a complete dumb with high knowledge. The only difference is that you can’t trust any knowledge from anywhere on the internet anymore.

You are contradicting yourself. The first half of your comment seems to deny the deterioration of knowledge. The second half admits that “you can’t trust any knowledge from the internet anymore.”

So which is it? Is our relationship with technology epistemically suboptimal or not?

There are a lot of reasons to hate AI, I don’t really buy into this one. “New technology only captures a subset of knowledge” is just not a new issue, and fearmongering around it is not new either.

I would be concerned about language collapse (i.e. models incrementally forget vocabulary + fail to adapt to new terms leading to language becoming incrementally worse) before I would worry about knowledge collapse, and I’m not that worries about language collapse.

Well, unlike you, experts are worried. Consider how knowledge is gathered, stored, and disseminated. AI interferes with each of these steps in dramatic ways.

Fortunately I have media literacy and know that the existence of someone marketing a scary narrative doesn’t actually provide any credibility. I can read scary headlines and pass it through the “does the make sense?” filter before getting worried.

I agree the friction + manipulation that LLMs introduce is troubling, but “knowledge collapse” is the extreme end of that concern.

Do you think Terrance Tao (who’s been sanguine on AI) and all the other top mathematicians on the planet succinctly explaining how LRMs are going to decimate our understanding of mathematics are marketing something?

You’re an ignoramus whose tiny dregs of insight can survive virtually cataclysm. I’m not so sure about humanity’s larger domain of shared knowledge.

“thing doesn’t work on me, therefore it’s the same for everyone or not a problem”.

Fuckin classic lol.

It’s wild when people like you act wise but expose the complete lack of worldview extremely fast.

I would be concerned about language collapse (i.e. models incrementally forget vocabulary + fail to adapt to new terms leading to language becoming incrementally worse) before I would worry about knowledge collapse

Why so?

The exact same mechanism is on display here. Ask an LLM for information, and it’s more likely to condense an answer, not display additional relevant information, and slowly but surely produce less and less outputs including that given information.

Sources then begin to be generated by the AI model that adds less and less of that information over time, LLMs cite those new sources that don’t include that information, etc, etc, and now that information is just not present in a majority of sources, and is never stated in any LLM responses.

If anything, this is more likely than models forgetting certain vocabulary, because while the underlying tokens for a lot of that vocabulary still exist in the model, the subsequent combinations for a given piece of information are substantially less likely, and the models’ finite size can only store a subset of that actual information in statistical averages. (i.e. you can store the alphabet and a dictionary of every word in the English language, and many simple phrases, in a tiny fraction of what the sum of Wikipedia is, and Wikipedia itself will contain a subset of all information on the internet, and so on, but you can still theoretically write any English content you’ll find on Wikipedia or the internet using that alphabet and dictionary)

It’s a pretty obvious issue to anybody who cares to look.

Not really an issue if the knowledge they expose you to is still more than you can absorb. Classic web search can show you 10x more info but you will just skim it anyway and absorb only tiny portion of it.

This isn’t about individuals. We safeguard knowledge institutionally. For instance, everyone should read this open letter from the top mathematicians on the planet, describing how we are about to undermine our mathematical understanding IDIOCRACY style. (I’m paraphrasing.)

Essentially, mathematical proofs are worthless without the communities that discover them. We care about math proofs because they create these communities (safeguarding knowledge for the next generation), not vice versa.

Having ChatGPT assert a proof we lack the expertise to check defeats the purpose. Not to mention, we already know it will get things wrong because of Gödel incompleteness so this whole situation is peak irony.

Yes but that’s a completely different issue than the one the article describes.

You’re right, but the article had a similar flavor. Specifically, when looking up information, users should be discovering curious details that raise additional questions. The AI is a homogenizing service that removes this “creative discovery” aspect of information gathering.

Really narrow view of knowledge.

The danger of catapulting an entire generation into ai-fed answers isn’t the amount or quality of knowledge, it’s training people to ask and NOT verify, a pillar of attaining wisdom and true expertise.

Yes but that’s a completely different issue than the one the article describes.

You keep saying that and it’s not true.

Knowledge collapse due to AI feeding everyone the same answer without people having to do real research, aka comparing sources, will absolutely happen.

It’s like people can’t see data and extrapolate anymore.

Can you not see the numerous ways that funneling all information through a single source like AI is a terrible idea?

Hell off the top of my head, the consolidation of information to AI would easily let whatever company/govt that’s in charge change a narrative on a scale humanity has rarely or never seen.

Yes but that’s a completely different issue than the one the article describes.

Not the fact that now every government can control that information if they wanted to?

What about that high encrypting though?

Dumb alarmists raise alarm because they don’t have a real job and are afraid they won’t get funding if they don’t create sensationalism every time