Opinion | Silicon Valley Is Investing in the Wrong A.I.

Jevin West and I have long argued that if OpenAI survives, it will be as 1000 squirrels in an LLM-shaped trenchcoat. A thin language layer will farm tasks out to 1000 domain-specific modules, much as Google evolved from PageRank to 1000 different systems.

Here's Gary Marcus (h/t @mehr.nz)

Federal civs swear the exact same oath of office as commissioned officers and I for one think the next dem admin needs to start holding them to it

Oh no

“Europeans who oppose migration” would get privileged access to move to the US.

I always have to chuckle at immigrants who oppose immigration, saying this as a (hopefully) self-aware repeat migrant.

Anyways, sounds like DEI for white racists.

How tf can these agents sleep at night after taking a young kids’ parent away

Holy crap. Trump will make the FCC crucify Jimmy Kimmel for this.

“I want to erode democracy when it prevents me from doing the (possibly quite dumb) things I want to do” is not a disturbing position one needs to seriously engage with and debate.

I didn't know Charlie Kirk existed until the early 2020s, when I learned about him and Turning Point USA while researching U.S. protest activity for my old job. From what I saw over the next several years, I came to detest him, and to see him as an important figure in the ascending American right.

Overpaid former SF now contractor without qualifications incoming

How are constitutional law folks feeling these days? (Genuine question)

*19 September

Or before the civil rights era a 100 years later, considering a large minority was disenfranchised in the South until the

E.g. the whole Iraq IED-heavy nature was dependent on a shallow but static occupation (somebody can emplace an IED without being caught and you’ll drive by it), so outside of those conditions you don’t really need to have the specific EW and MRAPs etc.

Has anybody written about the conditions under which the current drone-heavy quasi-stalemate is possible? E.g. if one side has air superiority and/or sufficient arty ammo, or better skill at maneuver, do you still see heavy drone use defensively, offensively?

Link to the book website: craftinginterpreters.com

A screenshot of the landing page for the online version of Crafting Interpreters by Robert Nystrom.

If you want to learn more about how programming languages work under the hood, can't recommend Crafting Interpreters by Robert Nystrom enough. Absolutely stunning and well-done book, and available for free (by the author) online. I just finished the first part on a Java-based tree-walk interpreter.

Charlie Kirk v @charliekirk11 Hillary For Prison 11:04 PM • Dec 21, 2015

Jason Campbell & @JasonSCampbell Charlie Kirk: "If Alexander Hamilton, Thomas Jefferson, George Washington, or John Adams saw drag queen story hour, they would mobilize the Minutemen" CHARLIE KIRK SHOW 0:54 291K views 10:51 AM • 8/10/21 • Twitter Web App 完

Charlie Kirk @ c@charliekirk11 • 16h Residents of Springfield, OH are reporting that Haitians are eating their family pets, another gift of the Biden-Harris mass immigration replacement plan. Liberals will soon be lecturing Americans on why they need to be sensitive to Haitian culture and accept this as the new Show more Springfield Ohio Cri... BIDEN FLOODS OHIO TOWN WITH 20,000 HAITIAN MIGRANTS: TEN TO A BEDROOM Warning to all about our beloved pets & those around us!! My neighbor informed me that her daughters friend had lost her cat. She checked pages, kennels, asked around, etc. One day she came home from work, as soon as she stepped out of her car, looked towards a neighbors house, where Haitians live, & saw her cat hanging from a branch, like you'd do a deer for butchering, & they were carving it up to eat. I've been told they are doing this to dogs, they have been doing it at snyder park with the ducks & geese, as I was told that last bit by on A. Moreno/Anadolu Agency via Getty Images, Pile Rangers & police. Please keep a close eve on these animala @ 2к 179.6K O 22K ill 2.4M 贝 Elon Musk & @elonmusk Follow Apparently, people's pet cats are being eaten

Here's Trump turnout chair, Charlie Kirk, condemning the women who "undermine their husbands" by voting for Harris: "It's just so nauseating where this wife ... who comes in with her sweet husband who probably works his tail off to make sure that she can go and, you know, have a nice life, provides for the family. And then she lies to him, saying, 'Oh yeah I'm going to vote for Trump' and then she votes for Kamala Harris as her little secret in the voting booth." GO -OTE NOW The Megyn Kelly Show SiriusXM

I’m choosing to remember Charlie Kirk’s contribution to American civic life in his own words.

Apropos of nothing, here's a thread of some Charlie Kirk highlights

PS: the title is totally hype BS, sorry

Clearly much better as it reduces the number of wrong answers, but still...half of the not i-don't-know answer are wrong. And I'm very skeptical that this is _the major_ cause of hallucinations. Rather, still seems to me like that is baked into the architecture.

I'm not a LLM expert, but that seems like a simple and compelling point to me.

What I'm less sure about is how much of hallucination is simply due to that. In the blog post they have some metrics, and 25% correct vs 25% wrong vs 50% "I don't know" seems like a stark tradeoff. (2/3)

Why language models hallucinate OpenAI’s new research explains why language models hallucinate. The findings show how improved evaluations can enhance AI reliability, honesty, and safety.

OpenAI has a new paper out, "Why language models hallucinate". I've only read the blog post and browsed the paper, but:

The core is a simple point: benchmarks reward you for guessing the correct answer rather than saying "I don't know", so confidently BS'ing gets you higher scores. (1/3)

Here is a list I post from time to time. My answer to a reader who asked me: what could journalists do NOW to break with some of their more corrosive habits. Defense of democracy seen as basic to the job. Symmetrical accounts of asymmetrical realities seen as malpractice. “Politics as strategic game” frame seen as low quality, downmarket, amateurish— and overmatched. Bad actors with a history of misinforming the public seen as unsuitable sources and unwelcome guests. Internalizing of the “liberal bias” critique seen as self-crippling, a historic mistake in need of correction.

Here's something I post from time to time. My answer to a reader who asked me: what could journalists do NOW to break with some of their more corrosive habits.

Love that they are explicitly calling a pro-democracy movement.

'No Kings' Unveils a Big New Trump Protest, and the Stakes Couldn't Be Higher Ezra Levin of Indivisible on how protest can stop Donald Trump’s authoritarianism.

The pro-democracy movement just announced its next big protest.

Read all about it:

The ‘Fortress Russia’ economy has adapted well to pressure. But stagflation presents an opportunity for the West Now could be the time to ratchet up sanctions pressure on Putin and change Moscow’s risk analysis of prolonging the war.

A balanced assessment of Russia's economy. "Full year real GDP growth forecasts are being revised down to less than 1% for 2025, but inflation expectations remain elevated...there is no end in sight for the current stagflationary macro situation." www.chathamhouse.org/2025/09/fort...

White House announces that Stephen Miller has taken residence in a warren of underground tunnels he burrowed under the new Rose Garden patio.

Abstract: Under the banner of progress, products have been uncritically adopted or even imposed on users — in past centuries with tobacco and combustion engines, and in the 21st with social media. For these collective blunders, we now regret our involvement or apathy as scientists, and society struggles to put the genie back in the bottle. Currently, we are similarly entangled with artificial intelligence (AI) technology. For example, software updates are rolled out seamlessly and non-consensually, Microsoft Office is bundled with chatbots, and we, our students, and our employers have had no say, as it is not considered a valid position to reject AI technologies in our teaching and research. This is why in June 2025, we co-authored an Open Letter calling on our employers to reverse and rethink their stance on uncritically adopting AI technologies. In this position piece, we expound on why universities must take their role seriously toa) counter the technology industry’s marketing, hype, and harm; and to b) safeguard higher education, critical thinking, expertise, academic freedom, and scientific integrity. We include pointers to relevant work to further inform our colleagues.

Figure 1. A cartoon set theoretic view on various terms (see Table 1) used when discussing the superset AI (black outline, hatched background): LLMs are in orange; ANNs are in magenta; generative models are in blue; and finally, chatbots are in green. Where these intersect, the colours reflect that, e.g. generative adversarial network (GAN) and Boltzmann machine (BM) models are in the purple subset because they are both generative and ANNs. In the case of proprietary closed source models, e.g. OpenAI’s ChatGPT and Apple’s Siri, we cannot verify their implementation and so academics can only make educated guesses (cf. Dingemanse 2025). Undefined terms used above: BERT (Devlin et al. 2019); AlexNet (Krizhevsky et al. 2017); A.L.I.C.E. (Wallace 2009); ELIZA (Weizenbaum 1966); Jabberwacky (Twist 2003); linear discriminant analysis (LDA); quadratic discriminant analysis (QDA).

Table 1. Below some of the typical terminological disarray is untangled. Importantly, none of these terms are orthogonal nor do they exclusively pick out the types of products we may wish to critique or proscribe.

Protecting the Ecosystem of Human Knowledge: Five Principles

Finally! 🤩 Our position piece: Against the Uncritical Adoption of 'AI' Technologies in Academia:
doi.org/10.5281/zeno...

We unpick the tech industry’s marketing, hype, & harm; and we argue for safeguarding higher education, critical
thinking, expertise, academic freedom, & scientific integrity.
1/n

#ANewHope 🙏