if we had a word for it, probably should have been using it to describe crypto and meme stocks for a while

Would also distinguish between "financialization" and actually *financing* something. The financial system finances things. Businesses. Buildings.

This finances nothing. There is no potential for any downstream productivity. It's just a money and attention suck that by definition produces nothing.

"What if we could create an entirely new asset class that could suck up unlimited capital and was guaranteed to never produce anything of value for anyone?"

Absolutely!

But it's nice to have the option :)

Also has convenient luggage luckers and a walk down to the main square that unfolds right from the station.

One of the best cities you can squeeze in as an "on the way" tourist stop in all of europe.

Unless the court-proof scheme is just a series of schemes with different rationale, each taking a year to litigate.

Probably not at the moment (if they are like Claude, office suite ingestion is pretty sus), but yeah that would give it a pathway to have a better shot at it.

The "callout" on the chart is just flatly wrong, though. It points to 600/100k as the moment >1% of the pop is in prison and, moreover, the graph *never* shows the line exceed that level.

Shows how awesome these things can be at remixing the unambiguous, but faceplant on the simplest math.

>he hates chess because it has no fog of war.

and it was in comparison to Polytopia, a game for children that can be beaten completely beaten in a couple of days

We make a big deal about "r's in strawberry", but that's an understandable flaw given that it sees words as vectors.

We should be *much* more concerned about the "9.11 is greater than 9.9".

It can see those numbers just fine, and SOTA models were screwing up the *most* basic math just a year ago.

So, paired with this, it's him?

bsky.app/profile/fint...

relatable tbh

"The extra 'S' is for 'SSS'"

"What's that extra S for?"

"That's a typo"

Absolutely. Even if they *could* eventually do the work of purpose built algorithms for hard problems, it would both be (1) *insanely* inefficient and (2) non-debuggable.

Even aiming to have them write sophisticated tools in place is far better than hoping to get that stuff in the model itself.

NLP and Computer Vision were/are separate things to which Machine Learning can be applied (and yes all of it was/is AI).

Neural Nets have been a core Machine Learning technique from the beginning, just one of several techniques for interpreting a feature space (and now the winner).

It is ML! All of it! It's a very broad category.

For decades, Neural Nets were just one type of ML, which was one sub-field of AI.

But AI is still reasoning and optimization fields (which LLMs do poorly!), they just get 0% of the attention.

>"Does that make sense?"

"Oh, yeah, totally" <rapidly mashes red button under table>

Hegseth - like his boss - doesn't seem to think through second order effects all that clearly.

oh come on does lizza's story about nuzzi really have a main character named izzy?