dlkf

What country are you referring to? The American middle class is objectively very wealthy: https://www.noahpinion.blog/p/no-the-us-is-not-a-poor-societ...

I mean, the matrices obviously change during training. I take it your point is that LLMs are trained once and then frozen, whereas humans continuously learn and adapt to their environment. I agree that this is a critical distinction. But it has nothing to do with “meaningful internal structure.”

> Previous generations of neural nets were kind of useless. Spotify ended up replacing their machine learning recommender with a simple system that would just recommend tracks that power listeners had already discovered.

“Previous generations of cars were useless because one guy rode a bike to work.” Pre-transformer neural nets were obviously useful. CNNs and RNNs were SOTA in most vision and audio processing tasks.

> Current AI systems have no internal structure that relates meaningfully to their functionality

In what sense is the relationship between neurons and human function more “meaningful” than the relationship between matrices and LLM function?

You’re correct that LLMs are probably a dead end with respect to AGI, but this is completely the wrong reason.

Now do drug-development.

If you take a vote of 10 random people, then as long as their errors are not perfectly correlated, you’ll do better than asking one person.

https://en.m.wikipedia.org/wiki/Ensemble_learning

If we did this fifty years ago, we’d be having this discussion by snail mail.

There is no epistemological collapse. Access to accurate information has never been so fast nor so easy. To be sure, lies are spread on the internet - but people believed all sorts of bullshit before the internet. Those who want to claim there is a crisis don’t have a principled argument as to how things are worse.

I implore everyone reading this to google the Sokal hoax before decide whether these guys are worthwhile.

Concrete Island by JG Ballard

Libra by Don Delilo

Deep Water by Patricia Highsmith

Assuming you value 10 days at the cost of 1-2 AC units.

The main difference is that the media report will be read.

Would new developments (or old developments that had gone underappreciated) in an adjacent stem field count as ”added cultural context” ?

A neat list, but there is a bias toward authors who published one controversial (and not necessarily good) argument that got a ton of rebuttals. Williamson is the worst offender.

The article is rife with references to a mediocre tv show, but it doesn’t contain a single example of the principle as applied to a firm. The author is an expert in something, but it isn’t business.

It’s interesting to consider whether a simulated rainstorm is in fact possible. Not a crude numerical simulation like those used for forecasting, but one fine-grained enough to accurately predict the trajectory of every drop.

In the regions where it works (PNW, Quebec, etc) we could easily build more. The hurdles are regulatory. The regulation isn’t baseless - a dam will affect the local ecosystem adversely. But that’s a tradeoff we choose rather than a fundamental limitation.

Why is green in scare-quotes?

> It means that google doesn't have to aquire-hire your team, they can just poach all of them wholesale.

Imagine a skilled worker taking a new job that pays better. What a nightmare scenario.

Does the page 404 for anyone else? Is this a Europe thing?

> Literally any definition of something infinite can always be reduced to a procedure that recursively transforms or observes some prior state.

You can’t generate R this way. This is a consequence of Cantor’s proof.

Your critique of the use of ”bigger” in your strawman applies equally to your own use of ”longer” - neither loop will ever terminate. So in what sense is either one longer?

The diagonal proof is arguing that there is no bijection from N to {0, 1}∞ , despite the fact that both are infinite. The sense in which the latter is ”bigger” is that there are always elements left over that are covered by N.

I’m not sure that you can. What I’ve read is that in a lot of fields, the community is small enough that you have a reasonable idea whose work you are reviewing even if the author’s name is redacted.

It’s debatable to what degree ”attention” in LLMs relates to ”attention” in psychology. See Cosma Shalizi’s note on this http://bactra.org/notebooks/nn-attention-and-transformers.ht...

I’m not accusing Whitehead of the fallacy, I’m talking about the parent post.

Parent is suggesting that if we don’t take Deleuze seriously, we should also not take Whitehead seriously. (Despite Whitehead being a fav of the analytic crowd.)

But the Deleuze comparison, to the extent that it makes any sense, only applies to the later Whitehead. We don’t have to reject the Pricipia. The obvious position is that Deleuze is not-even-wrong, process philosophy is not-even-wrong, and the principia is merely wrong (which is amazing in philosophy.)

Motte (principia) and bailey (process philosophy).

Can we please stop stating things that are obviously false? One can easily make a good case that software engineers are not oppressed without lying.

I think this is actually pretty well-established, eg https://news.crunchbase.com/layoffs/analysis-big-tech-pandem...

Hume wrote morality is better felt than reasoned. There is a large tradition in virtue ethics following Hume (eg Bernard Williams) who defend our moral intuitions, our prioritization of family, etc.

> I know it's easy to ridicule the sheer stupidity the board acted with (and justifiably so), but take a moment to think of the other side. If you truly believed that Superhuman AI was near, and it could act with malice, won't you try to slow things down a bit?

The real ”sheer stupidity” is this very belief.