senderista

There are lots of ridiculous unenforced laws like that; it doesn't tell you much.

Not uncommon Jewish names.

Sol 6.1 is definitely not comparable to Opus 5.5 for me.

It is already the same technique behind abliteration: identifying the direction associated with an observed behavior (refusal, for abliteration) and projecting it out.

Musk isn't Jewish, if that's what you're insinuating.

"a library of confusion" is exactly what you should expect from a 100% vibecoded product as complex as a database.

Don't you ever merge commits that wreck CI and are too complex to fix immediately? That happens here all the time.

It's similar with AI-generated writing: AI can't transcend mediocrity, but it can elevate terrible writing or terrible code to be merely mediocre.

Well support teams are even closer to the customer than product and I think they would advocate for perf/stability over more features.

Lesson for every infra provider: physically separate your free tier infra from your paid tier infra.

Not even close. Foragers 10K years ago had better longevity and nutrition than medieval peasants.

Is the problem that pension funds wouldn't be solvent given historical index fund returns, so they're looking for something magical that promises to outperform the market?

Sorry, you're right, that was a knee-jerk response on my part.

"Examples of unacceptable behavior by participants include...discussing this Code of Conduct."

There's a great anecdote about that in the book _Dealers of Lightning_, where the Xerox execs were indifferent to a demo of the Star personal computer but their wives were ecstatic about it.

Indeed, it makes it harder to block addictive sites like this one in HOSTS :)

Filling in well-known gaps like this is the kind of thing the AI companies should be doing, not trying to scoop research in progress.

Everyone seems to agree that PE is bad, but what about pension funds and other institutional investors in PE, would they be insolvent without PE?

Can't wait for the hyper-realistic customer service bots and deepfake propaganda videos.

We seem to still be a long way from an AI equivalent of Grothendieck (i.e., creative system-builder). Maybe AI is close to replacing a problem-solver like Erdos.

About the only general lesson I can extract so far from these announcements is that apparently we underestimated the extent to which major open problems are amenable to known approaches (possibly from very disparate fields), if they are combined judiciously by a virtually omniscient search engine. Frankly it's a little disappointing to me that these problems didn't require major new creative advances (e.g. the invention of new subfields) for their solution. And arguably that's a loss, if humans, with their inferior search and recall capabilities, would have been forced to resort to more creative solutions.

Was not expecting to see some LLM tells in Murat's writing, although it's clearly been heavily hand-edited.

"Knowing the answer" means more than knowing the value of a single yes/no bit.

The neocons at The Bulwark are secretly rejoicing about the war with Iran.

Well they're not always very nice to their Arab citizens either.

I think that's just because both parties managed to keep demagogues from being nominated until 2016.

> The entire American system of checks and balances is not designed to get optimal leadership, it is designed to limit damage from bad leadership.

Well how's that working out? There's a reason the rest of the developed democratic world hasn't adopted a strong executive.

This has been widely discredited.

https://www.bbc.com/news/articles/c9y7l0pm8jkko

The value of math does not consist in providing yes/no answers to conjectures.

What about frequent in-class proctored quizzes?