Friday: Hili dialogue

Welcome to Friday, September 11, 2026, and the 25th anniversary of the 9/11 terrorist attacks, in which 19 people hijacked four aircraft, two hitting the World Trade Center towers and bringing them down, cone rashing into the Pentagon and the last, after passengers attacked the terrorist pilots of United 93, crashed into the ground.  There were 2,977 innocent victims on top of the 19 terrorists.

Here’s a video of the second plane hitting the South Tower while the first is on fire, and then both buildings collapsing. You may not want to watch it if it evokes strong feelings:

It’s also National Emergency Responders Day, National Hot Cross Bun Day, and National Women’s Baseball Day.

Rosh Hashanah, the Jewish New Year, begins at sundown tonight and extends to sundown on Sunday.

Readers are welcome to mark notable events, births, or deaths on this day by consulting the September 11 Wikipedia page.

Da Nooz:

*An article in the Free Press by Daniel Kokotajlo, who worked at OpenAI, says that AI may really destroy all of humanity. I did not take such scenarios seriously until I read this article and found that this possibility is openly discussed in AI firms.

On Tuesday, AI researcher Jacob Coxon announced his resignation from Anthropic in a mega-viral post on X: “I spent the last three years doing pretraining research at both OpenAI and Anthropic,” he began. (Pretraining is an important part of the training process for AI models like Claude and GPT-6.)

He continued: “Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. . . . These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources. . . . The people building AI earnestly believe that it could kill us all by the end of the decade.”

He’s right. I know this because I worked at OpenAI for two years, forecasting AI progress internally and working on model evaluations and alignment research, and because I still talk with friends at OpenAI and Anthropic.

But you don’t need insider knowledge to understand where Jacob and I are coming from. The argument is straightforward:

1. Anthropic and OpenAI are racing to train self-improving AIs, ultimately superintelligent ones—meaning models that exceed the cognitive capacity of humans—as fast as they can. (This is not a secret. They talk about this in interviews and on their websites.)
2. If they actually succeed at training self-improving, superintelligent AIs, that would be a reckless gamble with our lives. (This part is fairly obvious, but has been explained at length by many people over the years, including the world’s two most-cited living AI scientists.)
3. They may actually succeed soon.

Many people doubt the third premise. AI companies have an incentive to overhype their products. After all, it’s common for tech companies to overpromise and underdeliver, so why should we trust their forecasts?

To that I say: We shouldn’t. Instead, we should look at publicly available evidence and forecasts made by independent experts. Unfortunately, that evidence base supports the third premise.

. . .The recent rogue AI cyberattack—in which AI agents in a controlled OpenAI experiment escaped of their own volition and hacked the company Hugging Face—illustrates these realities. OpenAI tried to spin the incident in a self-serving way. (Go watch their presentation at the Black Hat USA conference; they basically use the cyberattack as a reason to encourage customers to buy their AI protection services.)

But should we therefore conclude that the whole thing was a setup, a conspiracy by OpenAI to make their product seem scary and therefore powerful? No. OpenAI didn’t want their models to attack another company. It happened anyway, due to inadequate alignment and security practices at OpenAI. And the Hugging Face affair isn’t an anomaly. Smaller-scale incidents have been happening at other companies (and third-party evaluators) for months.

. . .Once the frontier AI companies have fully automated the AI research process, things will start moving very quickly. AI will get rapidly smarter than us in a variety of different ways, oversight of AI will be itself delegated to the models, and it will be all too easy for things to go wrong in a way that we can’t recover from.

The U.S. military is planning to integrate AI into many aspects of its operations. Businesses are planning to use AI for many things, including to design new technologies and run robot factories that build more robot factories. Superintelligent AI wouldn’t even need to take over by force; the current plan is essentially to put it in charge.

If Kokotajlo thinks it’s possible, well, it’s possible, as I know almost nothing about this. But some human malefactor would have to be behind a wholesale attack on humanity.  Kokitajlo’s solution is transparency and oversight forced upon AI companies, for “These companies cannot control their own AIs today, and should not be trusted to do so in the future.” All of us are ignorant about this, but if the doomsayers are right, we should at least pay attention.

*Mirabile dictu (again): the Supreme Court upheld a Missouri Supreme Court decision invalidating a redistricting of the state that favored Republicans.

Holding a general election with a different map than what was used in the primary is rare legal territory, with no specific guidance so far from the courts on how nominees should be selected.

Thursday’s ruling landed just as a hearing about remapping was about to begin at the Missouri Supreme Court in Jefferson City.

That court’s judges were considering whether to hold the secretary of state, Denny Hoskins, in contempt for not following their order to use the previous House map, which would position Democrats to retain a seat in the Kansas City area.

Before Thursday’s order from Washington, Missouri Republican leaders had been insisting they would stick with their preferred map for the general election.

In Jefferson City, a lawyer from the Missouri attorney general’s office, William J. Seidleck, said the secretary of state’s office was “evaluating next steps.”

*There’s a new report in our student newspaper, the Chicago Maroon, noting that the proportion of males admitted to the school is at its highest rate in decades: 58%. (I would have thought that the proportion of women (42%) would exceed that of men. Further, the proportion of black and Hispanic students has dropped pretty substantially since 2022, with black admissions being halved: from 10% to 5%.

The share of incoming students in the College identifying as Black fell for the third consecutive year in academic year 2025–26, while the proportion who were male increased to its highest level in decades, according to detailed UChicago enrollment data released last week.

The University reports data on its first-year classes annually in the Common Data Set (CDS), a standardized survey that asks institutions nationwide to disclose student demographics, high school academic performance, and financial aid awards, among other statistics.

The College enrolled just 71 Black first-year students last year, down from 97 in 2024 and over 150 three years ago, the University reported in the CDS. The number of white students showed a corresponding slight increase, while enrollment among other demographic groups stayed roughly unchanged year over year. Black students made up about five percent of domestic students of known race in the Class of 2029.

. . .The CDS release comes two admissions cycles after the Supreme Court ruled in 2023 that colleges and universities could not explicitly factor applicants’ race into their admissions decisions, effectively banning affirmative action policies favoring historically disadvantaged demographic groups. The University had submitted a brief in the case opposing an affirmative action ban.

Without affirmative action, many selective universities have admitted fewer Black and Hispanic students and more white and Asian students, though those shifts have been uneven across institutions. At UChicago, white and Asian shares of domestic first-time enrollment have climbed by about three percentage points each since the decision took effect, while Black and Hispanic enrollment shares have dropped by three and four points respectively.

The number of new first-years who did not report their race to the University jumped sharply last year, from 52 to 126, or seven percent of the Class of 2029.

You can see the change for race below; the vertical line marks the legal prohibition of race-based admissions:

The College also continued a years-long climb in male enrollment, the CDS release showed: more than 58 percent of the Class of 2029 identified as male, a three-point jump from the previous year and up from 52 percent for the Class of 2025. That weight toward male students persisted even as the College reported nearly 5,000 more female applicants than male ones.

UChicago had not welcomed an incoming undergraduate class with as high a proportion of male students since fall 1987, according to National Center for Education Statistics (NCES) data.

The University is an outlier when it comes to gender breakdown. Nationally, women comprised 57 percent of undergraduates in fall 2022. The only other Ivy Plus institutions with more male than female first-year enrollees last year—the Massachusetts Institute of Technology (MIT), Stanford University, and Dartmouth College—also had more male than female applicants. Even so, none had a higher share of male first-year enrollees than UChicago.

The decline in Hispanic and black students is surely connected with the Supreme Court ruling, though it can be circumvented in various ways. But schools are afraid of being sued like Harvard was. The increasing number of male applicants, and the increasing disparity between the sexes here is mysterious to me.  (Note that the Maroon says the numbers of “genders” is two, and what they really mean is sexes.)

*The British Museum has opened its exhibit of the Bayeux Tapestry, a project that’s been in the works a long time. I’ve longed to see this 70-meter embroidery of the Norman Conquest of England (dated to the 11th century), but never got to Normandy to see it.

Here, from the tapestry, is the death of King Harold of England, apparently struck in the eye by an arrow or spear. The legend says (Hic) Harold rex interfectus est, “[Here] King Harold was killed”. The King is apparently the unmounted figure in the middle, holding onto the spear or arrow. But some people say he’s the figure on the right, also apparently struck by a sword. We do not know.

Halley’s Comet also appears in the tapestry, and we know it showed up in 1066, the year Harold was crowned King of England but also died at the Battle of Hastings. Here’s the comet in the tapestry: people are looking at it with the caption, in Latin, “These men marvel at the star,” while Harold gets the news of the phenomenon.

Were I a Brit (or going to England), I’d do everything I could to see this exhibit. But who knows–it may be sold out until it closes.

*The WSJ reports that two people (Roland Fryer and Bill Ackman) are giving a million-dollar award to scholars who have the courage to tell the truth about “taboo” topics.

One of us (Mr. Fryer) released a study in 2016—later published in the Journal of Political Economy—finding that police aren’t more likely to shoot black Americans than white Americans, after accounting for the relevant context. Colleagues who read the draft warned him not to publish: “You’ll ruin your career.” The paper was denounced on social media within minutes of release, and Mr. Fryer needed armed protection for more than a month. Privately, scholars who read the study called it rigorous, honest work. Publicly, many of the same scholars said nothing. No single study settles a question this large—but silence isn’t how science advances.

That silence is the problem—a cultural and an economic one. When the professional cost of publishing an unwelcome finding exceeds the cost of suppressing it, rational scholars suppress it. The result is that social-science literature reflects what researchers are permitted to say rather than what they actually discover. The scope of the problem is difficult to measure, because its most important effects are invisible—though the data we have suggest it is getting worse. In a 2024 survey of more than 6,000 professors by the Foundation for Individual Rights and Expression, 35% said they had toned down their written work for fear of controversy—roughly four times the share of social scientists who admitted the same at the height of McCarthyism.

Mr. Fryer raised the alarm about suppression of inconvenient findings in a November 2024 Wall Street Journal essay. He called for something like a MacArthur Fellowship or an X Prize for academic truth-telling. The prize should be large enough to matter, prestigious enough to serve as a public credential for scholars who were right when it was costly to be right. Shortly after that piece ran, we found each other and decided to launch the Carob Trust Prize for Academic Courage.

The prize awards $1 million each to as many as five social scientists a year who have demonstrated intellectual independence, published findings that were attacked rather than answered, and been validated by the evidence—despite the professional cost. The selection criteria are designed to distinguish courage from contrarianism: Nominees must show a sustained commitment to following logic and evidence regardless of pressure, a willingness to ask questions others avoid, and work that has shifted academic debate, public discourse or policy—often despite being misread, mischaracterized or vilified at the time of publication.

The inaugural prize is limited to the social sciences; in future years we hope to broaden it to additional disciplines and to add a category for institutional leadership. Nominations are open through Nov. 1, and the first winners will be announced in early 2027.

Well, good for them.  The prize is enormous (up to $5 million per year, and must reflect the money or fundraising of billionaire Bill Ackman, as Roland Fryer is a professor of economics at Harvard. They emphasize in their editorial that the prize will not be partisan; it will reflect simply those who publish credible findings that are considered taboo. In the end, of course, this will largely involve findings demonized by the Left.

Meanwhile in Dobrzyn, Hili is ethnically cleansing her territory:

Hili: Let’s try to settle this diplomatically.
Kulka: What do you mean?
Hili: Go back to your home.

In Polish:

Hili: Spróbujmy to załatwić dyplomatycznie.
Kulka: Co masz na myśli?
Hili: Wracaj do swojego domu.

*******************

From Stacy:

nchanged: Unchanged:

From The Non Sequitur Assemblage of the World:

From Bad Spelling or Grammar on Signs and Notices:

From Masih: a mother talks about her daughter murdered and mutilated for protesting. The mother (sans hijab) vows to take revenge for her daughter. This is heartbreaking.

Translation from the Farsi:

What a nightmare these people lived through. What crimes did the Islamic Republic commit during those two days to cling to power, and the full extent of it remains hidden. Listen to the harrowing words of Javidnameh’s mother, Maedeh Moradi-Kian. The 22-year-old bride whose wedding was just one week past. On the 19th of Dey, when the people of Iran took to the streets in Qaemshahr for freedom, agents of the Islamic Republic fired a military bullet into this girl’s head. The little bride of Iran fell to the pavement. Her mother says that when they found Maedeh’s body, one of her hands had been severed with a machete, and her body was missing a hand. We still don’t know if they cut off her hand before shooting her in the head or after. What we do know is that this crime, this mutilation of the body of this young woman who had come to the streets empty-handed, was the work of Khamenei’s thugs. Her mother lit a candle by her daughter’s grave and said, Maedeh is my hero. I won’t find peace until I take revenge for her. Mother, we hear your voice. She is the hero of us all, and you are not alone. We will carry the story of this atrocity and Maedeh’s brutal killing to the entire world.

چه کابوسی را زندگی کردند این مردم. چه جنایاتی جمهوری اسلامی در آن دو روز مرتکب شد تا در قدرت بماند و ابعادش هنوز پنهان است.
صحبتهای تکان دهنده مادر جاویدنام مائده مرادی‌کیان را بشنوید.
نو عروس بیست و دو ساله که فقط یک هفته از عروسی‌اش گذشته بود.

نوزدهم دی، وقتی مردم ایران برای…

— Masih Alinejad (@AlinejadMasih) September 10, 2026

My repost (and riposte) to Steak-Umm, whose account Jeff Maurer touts as “The Standout Philosopher of Our Time”.  Bosh!

This is totally bogus. Which one you choose is determined by the laws of physics.

Steak-umm is a lousy philosopher. https://t.co/ypvzE0jUm9

— Jerry Coyne (@Evolutionistrue) September 10, 2026

From Luana; how DNA analysis helps us reconstruct human history:

Here’s a better map tbh using both Qpadm and G25 https://t.co/vmbM6uaABb pic.twitter.com/JUlvJIvVL6

— SuN (@SuN_BR0NZ3) September 8, 2026

A good and surprising video about suicide prevention posted by the Number Ten Cat:

Every 90 minutes, someone in the UK or Ireland dies by suicide.

Today is World Suicide Prevention Day. Suicide is not inevitable. You can interrupt someone’s suicidal thoughts. You could save a life.

I’ve shared this video before; it’s message remains as important and powerful…

— Larry the Cat (@Number10cat) September 10, 2026

From Malcolm: Cat yoga!

The cat 🤣🤣🤣 pic.twitter.com/jnhIxp0X3t

— Rob (@_ROB_29) September 8, 2026

One from my feed; and a sad one:

🕊 The missing Twin Towers returned to New York for a few minutes

A light show was held on the eve of the 25th anniversary of the September 11 attacks.

The drones first rose into the night sky in a spiral, then formed the outlines of the destroyed World Trade Center towers.… pic.twitter.com/obSuNKuAX0

— NEXTA (@nexta_tv) September 10, 2026

One I reposted from The Auschwitz Memorial:

This Dutch Jewish boy was gassed as soon as he arrived in Auchwitz. He was four years old and would be 89 today had he lived. https://t.co/3K68Fmx3Xb

— Jerry Coyne (@Evolutionistrue) September 11, 2026

Two from Dr. Cobb. First, a cat “tattoo” that may be AI. Who knows?

Le chat a un tatouage de son ex.

— Johnney (@johnneygirard.bsky.social) 2026-09-10T01:24:13.636Z

. . . and the world’s tallest horse. He’s a big ‘un!

everyone, meet Hugowww.bbc.co.uk/news/article…

— The Museum of English Rural Life (MERL) (@themerl.bsky.social) 2026-09-10T13:54:20.024Z