it's not a statistical or a computational limitation, because that implies that if you have better statistical methods or bigger compute or bigger data you can get around these things the problem is that you're modeling word tokens; you can't get out from under that this isn't revelatory though

“You used to be able to say anything” comedians never been more quiet in their lives

This is the second time today I’ve seen a post from a news org that says “from [larger number] to [smaller number].” This has to be an LLM thing, right? It’s too awkward for a person to have written.

I see the New Republic is using AI.

When Turkey purged their universities, "The 'to be expelled' lists were prepared by the university administrations. The few universities where massive dismissals did not take place were the ones who did not submit any 'to be expelled” lists." merip.org/2018/12/turk...

If you are on the TPUSA list of professors or are an otherwise visible left academic, I strongly recommend reviewing your digital footprint. Is your address public information? What about your family members? crimethinc.com/2020/08/26/d...

A lot of people who were panic mongering about a woke cultural revolution are conspicuously quiet about a youth activist denouncing a professor for wrong think and that professor being fired as part of a campaign of public shaming by political leaders.

A top 50 US university yesterday fired an instructor, a department chair, and a dean because a course made some reference to transgender people without the brief official course description in the catalog saying “trigger warning: gender.” In conclusion, those intolerant lefty kids, cancel culture.

I’m so happy I never signed up for that site.

A lot to say about this, but one thing is that the Professor in question was a lecturer, and thus did not have tenure. The destruction of academic freedom is inextricably linked to the loss of worker protection in the university.

This is not the AI boom creating a new type of work, this is a classic example of managers using automation to deskill old ones. Creative workers who once made a higher wage creating original art and writing are now paid in piecework to edit automated output. Bosses are using AI to cut labor costs.

My lab's been attacked by an anti-wind group and their lawyers. They threatened the science funding of my whole university, to shut me and my undergraduate research assistants up. NYT just covered it. 1/n www.nytimes.com/2025/08/25/c...

We have got to stop treating the most powerful and rich people in human history as the most innocent smol beans who only make oopsies with their tech products. There is *so much* public evidence that they know what they are doing and that they are doing it on purpose.

The fact that the thing we're calling artificial intelligence *can't do math* and yet we're jamming it into programs that successfully *have done math* for decades, then warning people against using the AI to do math, seems like an excellent summary of where we are.

Nvidia gets the spotlight, but *two* major groups are selling shovels in the AI gold rush, who profit whether or not AI products pan out: Chipmakers and Big Cloud—Google, Microsoft + Amazon. A new study by @davidthewid.bsky.social + Nathan Kim shows how the AI boom is fueling Big Cloud's land grab:

Also: “Grammarly is releasing a citation finder agent that automatically generates correctly formatted citations backing up claims in a piece of writing.” So not just formatting. But finding evidence to support your argument after you’ve made it and without you reading the cited source.

Oh, so we're just automating Rate My Professor now.

There is neither truth nor reason in markets, only quantifiable efficiency. The same seems to apply to AI: it is essentially an optimization mechanism. While optimization can be beneficial in clearly defined, socially embedded use cases, it is ultimately destructive when pursued as the sole value 3/

That's no coincidence. In fact, the first artificial neural network, Frank Rosenblatt's «Perceptron», was directly inspired by Friedrich von Hayek's neoliberal theories. For Hayek, both capitalist markets and the human brain were examples of spontaneous order without central control or planning 2/

In many respects, «artificial intelligence» is less a model of the human mind than a technical implementation of neoliberal market theory: a simulation of supposedly superhuman rationality optimized for efficient problem solving, which leads to mostly irrational and ultimately destructive results 1/

If you're just learning about that MIT retraction/distancing with the AI working paper and want to know more, this was a really good essay on it (and I did a mini thread where I tried to understand the media's role)

Necessary tools from the bad old days (or: the smallest taste of why marriage equality matters.)

You can see some collected info here. pasts-imperfect.ghost.io/pasts-imperf...

Say it again for those in the back: HUMANITIES USUALLY SUBSIDIZE STEM at most R1s. Yes we have receipts.

Anyone making this argument isn’t paying attention to what has been happening at Health and Human Services. And this thinking is a big part of how we got here.

Jordan Peterson is beside himself with glee.

Sunk-cost fallacy and cheating students are the only things keeping AI in business at this point. Just look at how usage of ChatGPT plummets as soon as the semester ends.

Every night at 4 am, my cat uses his litter box while meow-howling and digging like he’s trying to reach the earth’s core. During the day, quiet and efficient elimination. At night, scream pooping. It’s really unpleasant.

It's important to note here that we're not just talking about faculty & university staff. 100s of U.S. colleges are in towns or small cities where they form the major industry & lifeblood of the community. Loss of students -> loss of rent $, restaurant $, bars, bookstores, gaming stores...

If someone said my writing abounded “with interminable, convoluted sentences that teem with digressions and then awkwardly limp toward disorienting conclusions,” I would crawl in a hole and stay there. Just me quietly weeping in a deep hole forever.

This is the plan to try to sustain OpenAI while continues to set billions of dollars on fire. Put their AI models at the center of education to ensure a generation of people can’t function professionally without them.

Leah Belsky, a person leading Open AI's charge to foist its products on universities and schools is paid > $1M/year. Her net worth is > $25M. Before this, she ran (unsuccessful) efforts at Coursera. Ed-tech is fleecing public schools for private profit. www.nytimes.com/2025/06/07/t...

Weird how the strategy is to get everyone using it and then figure out the use cases. Seems inverted. 🤷🏿‍♂️

You can pry my em dashes out of my cold dead hands.

We drove from AZ to FL last week. And it’s surprising how many gas stations have instructional signage about appropriate toilet paper disposal.

This is gut-wrenching. It took decades, but the right has finally killed public media. And, like most government functions, most Americans won't realize how important it is until it's gone.

porting over my “banger novel openings” thread from the bird site

This isn't a list of jobs that A.I. can easily replace. It's a list of jobs that techno fascists don't think are worthwhile or meaningful and that they want to eliminate. They don't want humans to try to make sense of the world, they want us to believe whatever we're told.

I’ll add this: if you don’t want to read the literature in your field, brainstorm your own ideas, organize your own thoughts, and write your own arguments, you don’t want to be an academic. You just want to have a PhD.

I keep saying this. For a decade, people wrung their hands about “cancel culture” whenever campus communities objected to antisemitic white supremacist speakers. Universities were labeled the enemies of free speech. There was an all out moral panic. Now those same people are quiet.

We need to impress this upon every PhD student right now. Because it’s coming, and we don’t do anyone any favors by not warning them. (Especially with the job market - or lack there of - for the foreseeable future.)

A great thread about the problem with the “student as customer” model. The things students like (the class being entertaining and low-effort) don’t lead to good learning outcomes.

Kreitzer and Sweet-Cushman have a great summary of the research here. I am not joking in the least when I say it's easier to detect instructor gender than teaching or learning in evals. link.springer.com/article/10.1...

Evaluation is important and we have good ways of evaluating instructors (observations, audits, mentoring). But the evidence is clear that asking students to Yelp review their instructors the week before exams yields no valuable information on teaching or learning--just customer satisfaction.

If you work in higher ed, you need to get your folks together and end the use of course evals or at least bar their consideration in evaluation, promotion, and hiring. This needed to happen yesterday because they don't measure learning, they measure instructor gender, but now it's a snitch pool.

Friend: hey check out this good night Me: *already going gently into it* Friend: haha I don’t think you’re supposed to do that

We are going to have the fun thing happen again where a dozen cisgender female athletes discover they have Y chromosomes.

Stand by this: www.politico.com/newsletters/...

Trump and federal Republicans' $1.1 billion cut to public broadcasting will hit stations in AZ tribal communities especially hard, leaving them scrambling to fill budget gaps — & perhaps turning some parts of rural AZ into media deserts. buff.ly/UrZe7Zb

This is the behavior that I was describing as “VC Qanon” almost 5 years ago, for lack of a better term. And while Lewis has always been one of the worst, he’s far from alone. Just like Qanon, it encourages cult-like behaviors where paranoid thoughts are reinforced. www.anildash.com/2023/02/27/t...