iiinova — where your digital self keeps evolving

They didn't even provide a Lean proof for the Hilbert problem they claimed that they solved. But they did provide a disclaimer.

2282

Joe @twitskeptic.sirvin.com · 03/10/2026

Strange, that AI search apparently missed this chart:

1212

Joe @twitskeptic.sirvin.com · 01/10/2026

From my amateur point of view, the AlphaFold predictions they used in the paper for the full protein are pretty wild, but most of it is IDRs so, with the exception of the zinc finger domain, it's all very low confidence (warm colors are uncertain).

110

Joe @twitskeptic.sirvin.com · 01/10/2026

Oh, the “independent voice” first author works at OpenAI and graduated with some kind of bachelor’s degree two years ago.

000

Joe @twitskeptic.sirvin.com · 01/10/2026

OpenAI put up this “independent voice” commentary on the “AI” economy” and it’s page after page of inane AI slop. Just look at these figures! And, yes, they do talk about Dyson spheres. Can’t someone make it stop? Please? openai.com/index/the-et...

110

Joe @twitskeptic.sirvin.com · 29/09/2026

It's a very entertaining guilty pleasure to read. Someone should lock them all in a room and film a reality TV show where they discuss AI extinction risks.

120

Joe @twitskeptic.sirvin.com · 29/09/2026

Some pretty deep thinking in that report.

210

Joe @twitskeptic.sirvin.com · 27/09/2026

Yes, but maybe the results from humans are worse because there’s absolutely no incentive to put any effort into it.

070

Joe @twitskeptic.sirvin.com · 26/09/2026

And Marshal McLuhan in 1951 was concerned about media and advertising based manipulation generating "prolonged mental rutting". 3/

100

Joe @twitskeptic.sirvin.com · 26/09/2026

So familiar that T.S. Eliot wrote about cultural decline almost 80 years ago: 2/

100

Joe @twitskeptic.sirvin.com · 25/09/2026

Yep, that red dot is it. The only Flock camera in NW Seattle.

100

Joe @twitskeptic.sirvin.com · 21/09/2026

Renaissance man/founder contemplating late philosopher Daniel Dennett’s take on evolution during his recursive self improvement training.

020

Joe @twitskeptic.sirvin.com · 21/09/2026

This cyclical state transition diagram from W <-> Z is probably worth a billion $.

120

Joe @twitskeptic.sirvin.com · 21/09/2026

Nothing like a clear whiteboard to demonstrate that these startup bros are on to something and Cade Metz is on it. www.nytimes.com/2026/09/16/s...

120

Joe @twitskeptic.sirvin.com · 20/09/2026

Clickbait with an occasional moment of sanity

0111

Joe @twitskeptic.sirvin.com · 16/09/2026

Well, this is embarrassing

140

Joe @twitskeptic.sirvin.com · 15/09/2026

Another lesswrong/EA/rationalist tech bro who is full of himself. I miss the long gone fact checkers. bilalchughtai.co.uk/intellectual...

$AI Progress. 2024 was a big year for AI progress. AI is heating up in a big way, and as a result, I now feel much more urgency to contribute fast and put myself in positions of high influence. My AGI timelines continue to shorten, and I now think that within 3-5 years we will have systems that are capable of doing almost all human intellectual labour for a fraction of the cost (I like the phrase "labour replacing AI" for this idea, see here).$

000

Joe @twitskeptic.sirvin.com · 14/09/2026

The real existential AI threat.

Vinod Khosla, a billionaire Silicon Valley venture capitalist whose son has a health care A.I. company, has been particularly influential in conversations with top health department officials, according to people close to the matter. Those officials include Dr. Mehmet Oz, the head of Medicare and Medicaid. Chris Klomp, who faces a confirmation hearing this week to be the second-in-command at the health department, said on a podcast in May that the use of A.I. in patient care was the “holy grail” in the agency’s quest to improve the nation’s health and reduce costs.

Mr. Khosla has predicted for more than a decade that A.I. would largely replace doctors. “It’s over for doctors, human doctors,” he said at a July A.I. start-up event. “A.I. is just going to be better.”

In an interview, Mr. Khosla trumpeted Curai Health, his son’s company, which he has backed financially, saying it could offer an A.I. primary care provider that could give underserved people rapid access to care. “C.M.S. seems to be very excited,” Mr. Khosla said, referring to the Centers for Medicare and Medicaid Services.

020

Joe @twitskeptic.sirvin.com · 13/09/2026

Good summary of what a company that really was concerned about "existential risk" would do as far as cybersecurity goes.

First I would say use existing and well established industry veterans and companies with experience in cybersecurity, incident response, secure development lifecycle, policy, and both defensive and offensive capabilities.

Next I would make all evidence fully public including the user prompts, system prompts and everything an incident team would encounter or use in cases that go to court.

I would also propose training and evaluations requirements be met for those doing the work. This would be peer reviewed and agreed upon by a collective of industry experts and have realistic expectations and levels of knowledge.

All future scenarios where capability measurements are expressed should be independently verified by experts at all levels across all environments.

181

Joe @twitskeptic.sirvin.com · 12/09/2026

And then there's the obligatory "the US government should prevent competitors" statement.

$The main steps we can take to defend this gap are: Do not sell powerful AI chips or semiconductor manufacturing equipment to China, and crack down on chip smuggling operations and remote access to data centers outside China. Chips will be the main determinant of China’s AI strength. Crack down on unauthorized distillation by companies in authoritarian countries. Distillation of frontier models allows lagging companies to narrow the gap using a fraction of the cost it would take to develop their own AI independently. Strengthen security at the AI companies and prevent model weight theft.$

010

Joe @twitskeptic.sirvin.com · 10/09/2026

Oh yes, the same measured Kevin Roose who said this in late 2022 about GPT4. Great source.

050

Joe @twitskeptic.sirvin.com · 05/09/2026

Indie rocker Sean is on sale for only $27 as compared to $75 for physicist Sean for those who are interested in the cheaper Everettian branch.

000

Joe @twitskeptic.sirvin.com · 27/08/2026

FFS, that's not even the same study, unless Seattle, Victoria and New South Wales have been relocated to Sweden.

100

Joe @twitskeptic.sirvin.com · 26/08/2026

Might be time to shake up this committee and the Econ department.

Columbia College Announces 2021 Valedictorian and Salutatorian

Leopold Aschenbrenner CC’21, an economics and mathematics-statistics major from Berlin, and Sofia Montrone CC’21, a creative writing major from Los Angeles, have been named valedictorian and salutatorian, respectively, the top academic honors for the Columbia College Class of 2021.

Aschenbrenner and Montrone were selected by the faculty Committee on Honors, Awards and Prizes based on the strength, breadth, depth and rigor of their academic achievements, as well as on evidence of their intellectual promise, character and achievement outside the classroom.

Susan Elmes, the Andrew Barth Senior Lecturer in the Discipline of Economics, says of Aschenbrenner, “Leopold’s record of scholarship exceeds that of any student in the department in the last 20 years. His analytical brilliance is matched by his originality and creativity. Moreover, his willingness to engage in public conversations on ‘big’ topics marks him as a true public intellectual.”

010

Joe @twitskeptic.sirvin.com · 19/08/2026

I just don’t understand why this isn’t working out for him.

180

Joe @twitskeptic.sirvin.com · 16/08/2026

The credit scoring algorithm is ridiculous , but it doesn’t really matter if your credit score is >800. So don’t sweat it.

010

Joe @twitskeptic.sirvin.com · 16/08/2026

And, of course, Thoreau was making some of these same arguments about railroads over a hundred years ago. Yet we still have railroads.

Men think that it is essential that the Nation have commerce, and export ice, and talk through a telegraph, and ride thirty miles an hour, without a doubt, whether they do or not; but whether we should live like baboons or like men, is a little uncertain. If we do not get out sleepers, and forge rails, and devote days and nights to the work, but go to tinkering upon our lives to improve them, who will build railroads? And if railroads are not built, how shall we get to heaven in season? But if we stay at home and mind our business, who will want railroads? We do not ride on the railroad; it rides upon us. Did you ever think what those sleepers are that underlie the railroad? Each one is a man, an Irish-man, or a Yankee man. The rails are laid on them, and they are covered with sand, and the cars run smoothly over them. They are sound sleepers, I assure you. And every few years a new lot is laid down and run over; so that, if some have the pleasure of riding on a rail, others have the misfortune to be ridden upon. And when they run over a man that is walking in his sleep, a supernumerary sleeper in the wrong position, and wake him up, they suddenly stop the cars, and make a hue and cry about it, as if this were an exception. I am glad to know that it takes a gang of men for every five miles to keep the sleepers down and level in their beds as it is, for this is a sign that they may sometime get up again.

100

Joe @twitskeptic.sirvin.com · 15/08/2026

Pretty sure these guidelines will be violated within the first hour of deployment.

The G.L.O.V.E. SHALL NOT be deployed for any of the following:
• Verbal defiance or belligerence
• Punishment
• Torture
• Horse play

518316

Joe @twitskeptic.sirvin.com · 13/08/2026

That headline was so bad that even the NYT was willing to change it.

000

Joe @twitskeptic.sirvin.com · 11/08/2026

That article missed one of the most disturbing implications of the paper. It seems that all of these companies (OpenAI, Google, and Anthropic) only used one global key (one per company) to encrypt the detailed "reasoning" traces. This is inexcusable.

The consequences of this vulnerability extend far beyond intellectual property theft, representing a real-world privacy risk. Developers frequently share their session logs and encrypted thinking traces publicly
online, entirely unaware of the sensitive data hidden within the encrypted blocks. By scraping and
decoding 315,320 reasoning blocks from public repositories, we uncovered real data leaks, recovering 367
Personally Identifiable Information (PII) artifacts and 182 credentials; from genuine user sessions alone
these include 62 API keys, 33 passwords, and 30 personal emails. Alarmingly, in some cases, the recovered
PII did not even feature in the user’s input, having been injected invisibly from the model’s memory, or
it bypassed sanitization efforts because the user could not read the encrypted text before sharing it

190

Joe @twitskeptic.sirvin.com · 11/08/2026

Mark Zuckerberg, the people’s oligarch

100

Joe @twitskeptic.sirvin.com · 10/08/2026

Poor Zuck had to write a manifesto. He's always been dedicated to The People, and now we know that he's the most misunderstood man of the 21st century.

Meta is the company primarily focused on building personal superintelligence for everyone. Most other labs are focused on building AI for companies, governments, or other institutions, so if those labs lead, then the balance of power will favor larger institutions over individuals. Meta’s mission since our founding has focused on putting power in people’s hands. If our beliefs and principles lead, then the balance of power will favor individuals and a better future for everyone.

000

Joe @twitskeptic.sirvin.com · 06/08/2026

Salps are so inconvenient for people like this. www.whoi.edu/ocean-learni...

Salps go through two life phases in which they reproduce asexually, and later sexually, as hermaphrodites. In their asexual stage they produce long chains of identical salps connected together, which then break free and later reproduce sexually, with a baby salp growing up inside each parent in the chain. When food is plentiful, they can quickly create more chains, and each salp can increase rapidly in size. This superpower makes them one of the fastest-growing multicellular animals on Earth. Like all good things, the salp bloom comes to an end when all their available food is consumed.

000

Joe @twitskeptic.sirvin.com · 04/08/2026

Here’s an illegal Flock camera on NW Northwood Rd. Your fortune awaits you.

120

Joe @twitskeptic.sirvin.com · 27/07/2026

All the more reason for science communicators to be cautious when reporting phase I trial successes like the PDAC mutant KRAS vaccine. ~20% chance of progression to approval after phase I according to this: go.bio.org/rs/490-EHZ-9...

Table showing Phase II to approval for immuno-oncology ("IO") is 19.4%.

000

Joe @twitskeptic.sirvin.com · 25/07/2026

Check out these two figures from "physician-scientist" and AI-in-medicine booster Eric Topol based on a review he co-authored on aging clocks. The first shows a figure from the paper; the second shows the ChatGPT "simplified" version of same that he published on his substack. Looks credible to me.

Relative aging change from paper with bimodal peaks at 33 and 69 years

ChatGPT "simplified" version of previous figure with 3 peaks instead of two at around 35,60 and 80 with a hallucinated minima at 70 that isn't in the source image.

000

Joe @twitskeptic.sirvin.com · 21/07/2026

I think the bigger news from Nature is that Sanger sequencing was used for this GWAS.

From Nature, a headline about a "huge" GWAS study about borderline personality disorder is followed by an image from a Sanger sequencing run.

010

Joe @twitskeptic.sirvin.com · 17/07/2026

Hasn’t changed at all.

100

Joe @twitskeptic.sirvin.com · 30/06/2026

Gorsuch's waffling at the end of his own dissent is weird.

The government insists that aspect of the
order can survive any possible legal challenge, too, because
individuals can secure domicile in this country only if they
do so in compliance with federal law. See ante, at 57–58,
n. 10 (THOMAS, J., dissenting).
About that, however, I harbor doubts. Perhaps Wong
Kim Ark does not squarely foreclose the government’s posi-
tion. After all, that case addressed a child born to parents
who lawfully resided in this country. Still, I wonder: Is a
child born here to parents who have long chosen to make
this Nation their permanent home not a citizen under the
Fourteenth Amendment solely because his parents’ pres-
ence violates statutory law? If those parents are not domi-
ciled here, then where are they domiciled? And if the an-
swer is nowhere, how can we reconcile that conclusion with
this Court’s longstanding recognition that every person is
domiciled somewhere? See Desmare v. United States, 93
U. S. 605, 610 (1877). Because the executive order is not
facially invalid, these questions may not be properly before
us. But their answers are undeniably important to a Nation
committed to a view of citizenship open to all children born
here to parents who can call this country their home.

150