nyeah

In fact LLMs are trained to predict the next token in the training set. Of course sometimes a new text input doesn't match the training set, or it matches two or more places in the training set. LLMs use a neural network to interpolate, so that's fine. Please look this up if you have any doubts.

Ok. Now. I think you're adding something to the description above. Maybe what you're describing is something "emergent," or maybe it's basically just word vectors that were built in on purpose. You may be adding something correct, or something incorrect. Fine.

But it's not reasonable to say that the "reductive" description above is a "lie". It's not. It's more like a recipe. If you look at correct instructions for making steak, and you call the author a "liar" then you are missing something important.

Balancing parentheses requires semantic understanding?

Look, some folks are more impassioned about this stuff than I am. Maybe that's a good thing. But LLMs do in fact just try to predict the next token, using a very big training set. They're very impressive (at tasks the training set prepares them for). But that's how they work.

Sometimes. But bribery is also a crime, so using that word invites comment.

No, they were correct. In fact an LLM stitches together stuff it observed in its training data. That scales up way better than a lot of us expected, but it's still correct.

If you train it on lots of working code, then it's useful for coding. If you trained it primarily on non-working code it would produce nonsense.

Wallhack is locked into the DNA of CS for some reason. The most common defense by far is mass denial. For the great majority of players that seems to be good enough.

What's really thin is the oxide layer on the surface

https://en.wikipedia.org/wiki/Anodizing

When exposed to air at room temperature, or any other gas containing oxygen, pure aluminium self-passivates by forming a surface layer of amorphous aluminium oxide 2 to 3 nm thick,[4] which provides very effective protection against corrosion. Aluminium alloys typically form a thicker oxide layer, 5–15 nm thick, but tend to be more susceptible to corrosion.

Lines of code are a business liability. They are future cost.

In my opinion, you shouldn't have posted a false claim.

You claimed:

>crypto in general is a big improvment on banks

I responded to your claim. If you feel you can defend your claim, go ahead. If your only response is to change the subject, ok. That's your call.

I'll pick one incorrect claim out of many: Public ledger systems have no privacy at all, by design. You can hope nobody knows which wallet is yours, but that's not serious privacy. Today that's not even really hide-it-from-your-kid-brother privacy.

We know that it is much, much worse than the banks.

lol "self plagiarism". Max Planck got an "extra publication."

Counting papers is death. Everything connected with it is death. This is Max fucking Planck, who gave us the photon. We're judging him according to today's "standards." He's "failing."

Ok. So be it. We'll get what we incentivize.

>the idea of wave-packet collapse, when a measurement is done, is by itself completely unsound.

It sounds crazy if you read about the math first. Which almost everyone does ... now. But in the lab, if you set up a detector array that reports individual discrete particles arriving one at a time, that is exactly what you see. You just see random arrival times. You don't see a smooth wavefunction that then collapses.

The problem was never to start with the math and try to understand what's happening to a "wavefunction" or how the heck it could possibly "collapse." Instead they started with the crazy experimental results and made math that seems to always fit so far.

Maybe George Gilder is available. No PhD but lots of hands-on experience.

I think this is the real issue. Consumers love shiny cool stuff, but they don't like Clippy the paperclip. They like Siri when it helps but they don't like Siri when it impedes them.

What a conundrum! Why oh why are consumers reacting this way?

They've spent a lot of money on this technology, and what users want is just not that high a priority.

No. It doesn't have to be a bubble at all. It just has to be a cycle of rapid capital equipment build-out that returns to more normal levels in a few years.

Exactly. I'm glad someone else remembers the pre-eternal-September internet. (For me mid to late 1980s.)

Have you worked in a corporation? How sane was that corporation? Did it seem to even value its own survival? (Not corporations in general. In general they seem great. Just curious about the ones you actually did time in.)

Yeah this part scares me a little. I imagine it scares everyone who is more than a couple of years out of school. I hear that "the solution to LLM tech debt is more LLM." That might be true, but it might not be.

Thanks, it's very clear what you're saying.

Interesting data, thanks.

It's tuned to the audience. Bloomberg was traditionally for people who actually wanted information. People who were fallible and had limited knowledge.

Of course that mentality is obsolete. Now we all have infinite access to perfectly correct information via the internet.

Agreed. If the arguments seem sound and well supported, then all we can do is attack the tone.

That's possible, sure. But I think the answer is more likely in the numbers, not in just qualitatively saying AI isn't worth anything. Like if I pay $30k for an ounce of gold, I got value. Gold is worth something. But that amount of gold wasn't worth what I spent.

EDIT: In fact, parent comment has a link to some numbers.

[EDIT: Most] people don't want to go through the numbers. Ok. But there's a history here. When people don't want to see the numbers, certain kinds of things tend to happen.

Stop testing. If you test, the picture is only going to get worse.

This is a reasonable reaction to the title, but not to TFA.

Why listen to US progressives' opinions about Europe? Ask Europeans.

So much like Clippy.

Some might argue that lying to those who want to be lied to is still usually dishonest.