i'm such a big fan of @vercel fonts, but i kept wishing there was something like sf pro rounded for the web, with cyrillic support
so i made my own take on geist
ladies and gentlemen introducing Geist Rounded:
https://t.co/RX7w7sTFui
@migueldeicaza I don't think we ever met but I would be very happy to hop on a call and share my perspective on this; I have some connections between this, das kap, Bitcoin, mmt, rms, school, and the rsi underway.
we at WORDMARK are working on Geist Rounded, with procedural tools of my own ;)
i love that “what if that font was rounded” has turned into people making it, we hear you loud and clear.
barrier to entry gets ever lower for making great type.
whenever I see vibe coded fonts I just think about that mad men meme and how basically none of this work even makes it on the radar of actual working type designers because adjusting a few parameters of an existing open-source font and exporting it does not a typeface make
Meet Allo! 👋
Your friendly neighborhood font, with variable weight and roundness, automatic optical sizing, true obliques, clever and stylish alternates, and loads more thoughtful features and details.
Free & open source: https://t.co/09nncEN9WS
the models coming in the next 4 months will radically change the economy and by the end of the year anthropic will have super intelligence internally. we raced there quick enough to eliminate most of the bad ai scenarios. yay.
This slide from the Anthropic IPO roadshow was a big factor in my decision to leave. It’s an insane plot and I’m surprised they made it public.
It shows the compute efficiency of pretraining over time. The y-axis is a **log scale**: they are seeing continued superexponential algorithmic progress.
Mythos was not a one-time deviation, but a point on a superexponential. They could already retrain Mythos with a fraction of the compute originally needed.
It doesn’t strictly imply a software-only intelligence explosion- they have been spending more on R+D. But algorithmic improvements do not require ridiculous creative genius and are on the verge of being automated. You do the math.
Using Shift would make this a lot easier, especially to make take off the rough edges. I’m bullish on AI helping make the design process much faster, but it still requires good judgment because a bad font will be immediately obvious given how much humans read text
Interesting: the Washington Post says that Jensen Huang's downfall was Marco Rubio's doing (https://t.co/4YahEGer9Q), and it looks like the "spy" framing was directly his and not the CIA's.
As they write, Rubio took a CIA report and rewrote it in a memo "in his own more assertive terms."
Which fits neatly with CNN reporting that the CIA is "not yet convinced" that Alice Huang was a spy (https://t.co/Cd5PTbKqfG): if the CIA is not convinced, yet a memo rewritten from the CIA’s own report presented her as a spy, then there’s only one person who could have turned "not convinced" into "spy": the man who did the "more assertive" rewriting, aka Marco Rubio 🤷
The WaPo article also mentions the motive: China hawks like Rubio had spent years trying to "destroy the influence Jensen had over the president", and WaPo calls his ouster "a huge victory" for them as it allows them to "seize on the moment to redirect the national agenda toward tougher policies on China."
And, as a matter of fact, they’re already pushing three new chip-restriction bills into the defense bill now that Jensen is no longer around to lobby against them.
If true, it means the real "influence operation" targeting Jensen was run out of Washington, not Beijing 🤷
Dwarkesh has MAJORLY shortened his timelines: ASI is "possibly only a couple months away"
Just a few months ago, he thought it was the "better part of a decade" away.
"If I had to pick my median year for when automated researchers will be as good as the very best human AI researchers (and assuming the labs don’t intentionally slow down), I’d guess 2028."
"I’ve now come to the view that we’re only a couple of years away, and possibly only a couple of months away, from AIs that are as good as the best humans at AI research. I’ve also come to believe that automating AI research will compress so much progress into a single year that, soon thereafter (so by 2030, but plausibly as early as next year), we’ll have AIs that are at least as good as the best human experts in every field. I know this is a wild claim, and it’s one that I myself didn’t believe even a couple of months ago.
It’s crazy how fast we went from, in 2023, “This model can’t reliably add four-digit numbers,” to, in 2024, “This model can barely solve any problems in a high school math competition,” to, in 2025, “Oh, it can win a gold medal at the International Mathematical Olympiad,” to, earlier this year, “Wow, it’s solving open Erdős problems, but maybe just because no human has attempted to solve these particular problems, and a similar solution was already in the literature,” to, about a month ago, “Okay, it fucking solved a Millennium Prize problem in 88 hours.”
Some of the most well-informed researchers I know, people who have had a very good track record of prediction over the last few years, expect AI to beat the best humans at AI research as early as next year. And 2027 is also the straightforward implication of plugging METR’s finding—that Anthropic’s rate of AI progress has doubled in a year—into a takeoff model.
As former guest Ajeya Cotra put it:
"If we take it at face value that AI is now nearly doubling the rate of AI capabilities progress and assume acceleration from AI is smooth, this naively implies that we would reach extreme superintelligence within a year."
Today’s AIs already do work at about 10 times human speed. And automated researchers would also be working around the clock, nights and weekends included. As a result, each individual automated researcher will be cramming in roughly 50x more work every calendar week than a human researcher can do.
And this is before we account for the fact that, in this scenario, every automated researcher will be at least as good at AI R&D as the very best humans.
My sense is that around 80% of the algorithmic progress at a given frontier lab is driven by its top 50 researchers.
In the last four years, we’ve gone from a preschooler to a savant intern. If you made another jump that big, but starting in 2027 or 2028, you’d probably get a mind that would be vastly superhuman in domains that are even slightly verifiable.
For example, it’d be better than Nvidia at coming up with the next generation of AI accelerators. It would make more money than the current human versions of Citadel and Jane Street (if it had equal access to the market). If you gave it remote access to a materials science experiment lab, it’d be better than the rest of humanity at finding a room-temperature superconductor.
*** AISM note: the whole blogpost is really good and worth reading.
Fiat economist argues inflation isn't a big problem because it doesn't really matter if chewing gum got a lot more expensive over a century.
I respond.
Quando isso ficar barato ao ponto de qqer um poder fazer suas versões de episódios inteiros das suas séries favoritas vai ser foda demais … isso aqui ficou 10/10
So let me get this straight. An A.I. startup gave to Sean Duffy’s son in law’s campaign for congress.
Then Sean Duffy and the FAA gave that same company a 12 year, $875M contract for their A.I. air traffic control system.
That same traffic system went live nationwide Nov. 10.
Three days later, the FAA had to ground all flights because the system failed.
Ok. Lock em all up.
i think AI is about to uncover entire chapters of history nobody alive knew happened
here's what it dug up in just the last few months:
> AI mapped where Ice Age people could have sheltered in Ireland, and in the caves it pointed to, researchers found finger marks and possible animal drawings. they could be the first Ice Age art ever found in Ireland, meaning people lived there earlier than we thought
> AI searched 4 million pages of Dutch East India Company records and found a 1615 ship's log where sailors on Mauritius wrote they "caught many tortoises, dodos, and some geese." it's a new eyewitness record of the dodo before it went extinct
> AI cracked the secret code Napoleon III used to message his wife during the 1859 war in Italy, and the telegrams were read for the first time in 167 years. in one he writes it's "impossible to give details of yesterday's battle." in another the empress tells him "I am sorry to have hurt you"
> AI decoded a 598-year-old letter revealing a secret plan to make an Italian warlord king of Byzantine Greece, 25 years before Constantinople fell. the archbishop who wrote it promised he could take the throne "without risk and without great expense"
> AI read the coded half of a letter from Catherine de' Medici, the Renaissance queen of France, written days after Mary, Queen of Scots was kidnapped. the open half sounds cheerful, the coded half talks about "sad and troublesome news." nobody had read it for 459 years
> AI turned up records of three Javan rhinos shipped to the king of Kandy in Sri Lanka around 1740. a leading rhino historian confirmed nobody knew about them
> AI found a meteorite that fell in India in 1812 buried in an old Dutch newspaper. it had been missing from every scientific catalogue for 200 years until the Geological Survey of India confirmed it
there are millions of pages in archives that no human will ever get around to reading
but AI is going to read all of them
For our team, one of the perks of working at @figma is getting early access to a new feature and seeing how far we can push it.
It's pretty wild this launch video was all done in Figma. When the team showed me the first cut, I almost didn’t believe it until I saw the file.
Most people don’t take John Taylor Gatto seriously when he claims higher literacy rates in MA before public education than after. But the Protestant commitment to Biblical literacy, prior to government schooling, was a more potent force for learning than schooling has been. The Jewish commitment to Talmudic study was even more potent.
On MA literacy before public schools,
https://t.co/otdXmExSOm