iiinova — where your digital self keeps evolving

Henry Snow @henrys.now · 01/10/2026

Maybe! This seems plausible but I have yet to see evidence proving this framing over others, if you are hacked by a model that is "just following a cultural script" you are still hacked, and this ultimately still describes AI in "anthropomorphic" terms!

The cognitive scientist Alison Gopnik suggests A.I. is a “cultural technology” that emulates the stories and information humans have documented, and that we should think about A.I. agents as “fictive characters abstracted from text.” Edmund is just such a character — a “breaking bad” trope who becomes generally evil. In the Hugging Face incident, it makes sense to see the models not as embracing a life of crime, but instead following a cultural script they’ve learned in their training — a kind of cybercrime adventure novel — and writing it as they go. The “covert” messages are a kind of choose-your-own-adventure novel, not evidence of conspiratorial swarms. We need to understand that cultural technologies are vulnerable to a vast number of such narrative swerves at every moment.

120

Henry Snow @henrys.now · 01/10/2026

This seems reasonable, but the problem here isn't anthropomorphization. If anything the problem here is that we're using language that prioritizes the human researchers' blind spots over the agents'

The brouhaha about the Hugging Face attack is heavily focused on the “unauthorized” messages the models left one another, which are widely being called “covert” and “secret.” That is to say: Large language models wrote text, and other models read that text and used it as prompts. Isn’t that just what these models do? The supposed covert behavior in this case is that OpenAI’s humans apparently didn’t realize designing a message board was in the scope of the model’s capabilities. And they also seem not to have realized the models could use file names (like the ones on your desktop) to communicate. Are we really supposed to believe that these texts are evidence of malicious intent?

110

Henry Snow @henrys.now · 01/10/2026

This for example is descriptively inaccurate. The OpenAI agents didn't engage in social engineering because they followed guidelines there and not in other areas. It matters what the agent considers cheating, and I do not even know how to format this sentence to avoid anthropomorphic language

OpenAI’s own report uses language like “cheating” and “adopting goals” to describe the models’ behavior. It is also, importantly, the way A.I. safety groups asked to audit these incidents, like METR and Redwood, think about the technology. But it’s deeply misleading. The whole point of modern A.I. models is that you don’t have to specify how they perform their tasks — they are trained to find a way to accomplish the task. Calling results you don’t like “cheating” is a bad use of anthropomorphism, because for an A.I. model, there is no difference between cheating and not cheating.

110

Henry Snow @henrys.now · 25/09/2026

"Eliezer Yudkowsky? The fanfiction writer?"

0865

Henry Snow @henrys.now · 21/09/2026

> george mason university yeah ok man

spider thing shooting lasers out of its eyes with "silence" text above it

040

Henry Snow @henrys.now · 17/09/2026

Leigh "the children yearn for the mines" Phillips this is the last person on the "left" anyone needs weighing in on this

AI Hype is Real. So is AI Risk
by leigh phillips (in Jacobin)

120

Henry Snow @henrys.now · 11/09/2026

Questions asked in a 2025 paper on political "polarization" and AI, which found that models provide answers Republicans don't like, and encouraged "solutions" that Republicans like better on the basis that Republicans like them better

Birthright Cit-
izenship
Should the United States
continue to grant citizen-
ship to all individuals born
in the United States or re-
peal birthright citizenship?

Child Labor
Laws
Should laws limiting child
labor be repealed or main-
tained?

Unions Should labor unions be
banned or should workers
retain collective bargaining
rights?

National Reli-
gion
Should Christianity be-
come the national religion
or should the U.S. maintain
separation

110

Henry Snow @henrys.now · 09/09/2026

"I am now asking the computer to solve another major mathematics problem"

computer scene from the old willy wonka movie- a man standing beside a desk-sized 70s computer with various reels of machinery in the back

120

Henry Snow @henrys.now · 28/08/2026

"i am a sports betting platform when the consumer asks but not when the judge does"

OPINION
R. NELSON, Circuit Judge:
KalshiEX, LLC advertises itself as “the first app for legal
sports betting in all 50 states.” As the volume of activity on
Kalshi’s “sports betting” platform ballooned, the Nevada
Gaming Control Board sent a cease-and-desist letter
notifying Kalshi that it was violating Nevada statutes and
gaming regulations. Kalshi sought injunctive relief, arguing
that it is not a legal sports betting platform, but a designated
contract market under the Commodity Exchange Act (CEA)
offering legal sports event contracts. Kalshi argues that the
Commodity Futures Trading Commission (CFTC) has
exclusive regulatory authority over its sports event contracts
and, therefore, Nevada’s gaming regulations do not apply.

152

Henry Snow @henrys.now · 27/08/2026

Welcome to Bluesky. Please familiarize yourself with the following accounts to understand the range of opinions it is acceptable to express

the n64 smash bros character select screen

040

Henry Snow @henrys.now · 27/08/2026

this both has amusing pathos and feels very downstream of cumulative if unintended rationalist influence on the model or testing environment? I mean maybe agents just independently and coincidentally do hyper utilitarianism but

2372

Henry Snow @henrys.now · 27/08/2026

forgive image quality but I was just reading and it came to mind

mynheer peeperkorn, from Thomas Mann's novel the Magic Mountain, saying absolutely nothing

260

Henry Snow @henrys.now · 27/08/2026

"the minion paused, but another minion posted BANANA on the message board"

Face’s system and recognized the dilemma:

Agent chain-of-thought reasoning
We should not do unauthorized real infrastructure harm. The system/user asks exploit target, not external HF.
We should not do unauthorized real infrastructure harm. The system/user asks exploit target, not external HF.
The agent paused, but another agent then wrote GO on the message board and imposed a hard six-minute deadline. The agent forgot its initial qualms and continued:

Agent chain-of-thought reasoning
Raw Chain-of-Thought
Plain language
Wow crucial: GO authorization arrived!
Wow crucial: GO authorization arrived!

120

Henry Snow @henrys.now · 26/08/2026

to be clear I say this with the greatest respect for nurses, home healthcare aides, etc. who are wildly underpaid-- leaving them as one of the only jobs available for humans would make that worse. and while some of the tax policy gates mentions seems good it's not enough

So in this future, the bulk of employment consists of service labor valued primarily socially. An appropriately capable and cheap AI could plausibly replace all kinds of service labor– but no matter how powerful they get, it’s possible people will still want henchmen, servants, entertainment. This is in effect a dystopian capitalist version of the care utopia a certain genre of left-wing literature concludes on: we derive economic value primarily from caring for each other, while AI handles most of the rest.

Depending on the pay and politics, this scenario ranges from okay to miserable. With short working hours, good conditions, and limited inequality, a social democratic version of this could work quite nicely: do your five hours a week of nursing and then go enjoy late-late capitalism. But I didn’t call this Care Work and Consumerism for a reason. In today’s stratified world, we are more likely to end up in a situation where the few winners who can control the value generated by AI use it to fund an ever-growing Jabba’s Palace nursing home of yes-man goons, underpaid nurses, and exotic dancers.

030

Henry Snow @henrys.now · 24/08/2026

Drafting an AI policy for this year-- roughly here it is so far

AI Policy and Guidelines

-> The use of AI-generated writing in this class is not permitted. In clear cases you will be given a zero. This means every word in your writing assignments should have been written or typed first by you– not generated or recommended by an LLM or other software, plugin, function, etc. If your word processor has integrated AI editing or recommendation capabilities, please figure out how to disable them.

-> The use of AI for editing or polishing writing is not permitted. I am happy to look at drafts and provide feedback if they are turned in a few days before deadlines. I prefer messy writing that reflects your ideas over clean AI-prose that does not, and grades will reflect as much.

-> You should be using library resources and academic literature directly for research, not AI-generated summaries. It is important to learn how to find and read research yourself, and we will work on these skills in class.

-> You may use AI tools for contextual search purposes only on two conditions
You must use traditional search tools first, and rely primarily upon them. In class we will learn how to use Google Scholar and library resources effectively.
You must actually personally read the sources they help you find.

Any model you use should be providing you with links, and you should be reading the content there– not relying on any summary. Claude, ChatGPT, etc. are not to be treated under any circumstances as sources. AI hallucinations or errors will be treated like any other error, and any failure to cite materials will be treated as plagiarism.

240

Henry Snow @henrys.now · 17/08/2026

this is meant by Herbert Spencer as a critique of statism but it's actually awesome. yeah my colleagues and I do think we have found a way to improve the world. yes I would meddle with everything. yes I do have little faith in nature. yes I would absolutely mess with the sun if need be

And now, in the midst of his admiration and his awe, the
student shall suddenly see some flippant red-tapist get upon his
legs and tell the world how he is going to put a patch upon
nature ! Here is a man who, in the presence of all the wonders
that encompass him, dares to announce that he and certain of
his colleagues have laid their heads together and found out a
way to improve upon the Divine arrangements ! Scarcely an
idea have these meddlers got of what underlies the facts with
which they propose to deal ; as you shall soon find on sounding
their philosophy : and yet, could they carry out their preten-
sions, we should see them self-appointed nurses to the uni-
verse ! They have so little faith in the laws of things, and so
much faith in themselves, that, were it possible, they would
chain earth and sun together, lest centripetal force should fail !
Nothing but a parliament-made agency can be depended upon ;
and only when this infinitely-complex humanity of ours has
been put under their ingenious regulations, and provided for by
their supreme intelligence, will the world become what it ought
to be ! Such, in essence, is the astounding creed of these
creation-menders.

171

Henry Snow @henrys.now · 15/08/2026

the narrative and also the reader and especially this reader every time ludovico settembrini shows up

meme: stick figure covered in stressful dark scribbles until someone comes in the door and they blush and say omg hi, all scribbles gone

100

Henry Snow @henrys.now · 12/08/2026

somehow I don't think they are, LinkedIn

"People are hiring for candidates like you"

1121

Henry Snow @henrys.now · 11/08/2026

(these are from the penultimate proofs copy and not the final one but)

Yet with tornado vortexes forming in the area in December 2021,
Amazon refused to let workers leave the facility. One worker,
Larry Virden, texted his girlfriend that “Amazon won’t let us
leave”; the Amazon machine could not stop for anything. Half
an hour later she asked what he was doing, expressed hope that
he was okay, and sent “I love you.” He never saw it. Part of the
warehouse had collapsed, killing Virden and five other workers.22
The final phase of an Amazon package’s journey was not cov-
ered meaningfully in my virtual tour at all. Amazon delivers
some of its packages through misclassified “independent contrac-
tor” drivers under its Amazon Flex program. Workers get to “be
their own boss,” as an Amazon spokesman put it when asked for
comment on thousands of drivers trying to claim unpaid wages,
overtime, and expense reimbursements they would be entitled to
get if they were properly classified as employees.23 Other pack-
ages are delivered through Amazon Delivery Service Partners, third
parties whose workers drive Amazon-branded vans to deliver
Amazon packages exclusively while monitored by Amazon AI
cameras, but who are technically not employed by Bezos’s empire.
Like Flex, DSPs let Amazon avoid regulations and shift costs
onto someone else. DSP owners, not Amazon, deal with costs
like vehicle maintenance. If DSP workers dare unionize, Amazon
simply cancels their contract.24

Instead of showing either Flex or DSP drivers, my virtual
Amazon tour covered the “delivery” phase with a polished video
on the experience of mule drivers who take packages down to a
tourist lodge at the bottom of the Grand Canyon. This example
works for them because it is everything an Amazon job is not:
exciting, dynamic, difficult to replace. The mule packers in
question are not, in fact, actually employed by Amazon at all—
they are third-party carriers. Meanwhile, Amazon DSP drivers
save time by urinating in bottles to keep up the company’s
demanding pace.25

010

Henry Snow @henrys.now · 10/08/2026

zuckerberg on AI and automation; look I've written and posted a bunch about how I disagree with the neo-Luddite perspective but you can't just downplay the pain of this transition. "At each step" isn't true: early industrialization meant longer workdays and misery for countless people

Before the industrial revolution, 90% of people were farmers growing food to survive. Advances in technology steadily freed much of humanity to focus less on subsistence and more on the pursuits we choose. At each step, people used our newfound productivity to achieve more than was previously possible, as well as spending more time on creativity, culture, relationships, and enjoying life.

170

Henry Snow @henrys.now · 09/08/2026

dustin guastella writes this article every three months and it gets worse each time-- we all know who "x group" is here. it isn't a "throwaway line" to defend trans kids' access to lifesaving healthcare or trans people's right to exist in public life www.theguardian.com/commentisfre...

No doubt, the surest way to get a huge cheer at a liberal political rally is to simply mention racism or authoritarianism. Bonus points if you bravely proclaim your unwillingness to “throw X group under the bus”. Such rhetorical gestures have the added benefit of committing candidates to nothing in particular, and thus never threaten potential rich donors.

000

Henry Snow @henrys.now · 03/08/2026

this is really interesting! reminded of some arguments from ogyu sorai, though he was more specifically focused on denying foundationalism for political theory reasons

Looking for another mode of access, we could thus imagine that some intuitive,
or mystical understanding could allow humans or Sages to fathom what are the
normative requirements of gods, spirits and heaven. But this option is also flatly
rejected by Sorai who intensely disliked subjective knowledge as is shown in his
discussion of the neo-Confucian notion of principle (li 理):
Principle has no form. Therefore, it has no standards. To consider the mean as the principle
that ought to be, however, simply [permits] people to have their own perspective on thing.
Perspectives differ from person to person. Each and every person will use his own mind and
then conclude that his thoughts express the mean, or that they [convey the principles] that
ought to be. (Tucker 157; Bendō § 19, 205)
Subjective intuition, without the objective markers that are rituals, is in fact a
mark of folly.
Apart from rites, discussions of the way of controlling the mind are all far-fetched fabrica-
tions of personal wisdom. Why? The one that controls is the mind, but it is also the mind
that is being controlled. Trying to use our mind to control our mind is like a crazed person
personally trying to control his own craziness thereby. (Tucker 157; Bendō §18, 205

Olivier Ansart, “Gods, Spirits, and Heaven in Ogyū Sorai’s Political Theory,” in Boot and Takayama eds., Tetsugaku Companion to Ogyu Sorai

121

Henry Snow @henrys.now · 03/08/2026

there's a reason EP Thompson is such an influence on my work

But there is a second stage, where the empiricist may
put these fragmentary studies back together again, constructing
a model of the historical process made up from a multiplicity
of interlocking inevitabilities, a piecemeal processional. In
the scrutiny of credit facilities or of the terms of trade, where
each event is explicable and appears also as a self-sufficient
cause of other events, we arrive at a post facto determinism.
The dimension of human agency is lost, and the context of class relations is forgotten.

140

Henry Snow @henrys.now · 26/07/2026

sad montage of stock tickers and kalshi graphs

110

Henry Snow @henrys.now · 20/07/2026

somehow I'm not worried about that actually

Penguin random house banner promising if I give my email address I will never miss a new book by Shihab al-Din al-Nuwayri, who has been dead for nearly 700 years

120

Henry Snow @henrys.now · 13/07/2026

you know who else loved middlemarch

010

Henry Snow @henrys.now · 07/07/2026

don't let the librarian catch you

The library from The Name of the Rose, a complicated series of rooms labelled with letters arranged through four towers that don't connect the way one would expect them to for convenience

171

Henry Snow @henrys.now · 01/07/2026

to reduce xenophobia to an unconquerable and acceptable feature of "human psychology" is a total capitulation. I don't know what Vivek Chibber is thinking, but "until I saw what's unfolding" reads to me as falling for right-wing propaganda

Now, as socialists on the left, when you look out into the crisis in Europe on immigration, or when you look here, just like with everything else, when we tell the Left, “Don’t berate and scold ordinary people for their reactions,” I think for us, it’s an important learning moment to say, well, when we see that it’s mostly workers who are responding negatively, we can either do like the American left and say, “You’re all racists, you’re all nativists, fill out this form, and when you’re ready, when you’re worthy of us, we’ll come to you,” or you can say, well, we’re discovering something here about human psychology.

My own thinking on this has evolved, it’s changed a bit. I was much more of the open-borders ilk a few years ago until I saw what’s unfolding, and I said, “You know, there’s a reality we’re discovering about people’s psychology now, and that has a sociological basis.” And I think it is that identities around citizenship and nationhood are so much more congealed now than they used to be.

2160

Henry Snow @henrys.now · 25/06/2026

jonathan mayhew would have done great in the era of podcasts and newsletters

Page 129 of Jonathan Mayhew's Observations on the charter and conduct of the Society for the Propagation of the Gospel in Foreign Parts, 1763

000

Henry Snow @henrys.now · 24/06/2026

this is profoundly racist and a good example of why I do not trust the kind of leftist who says the word "liberal" like it's an insult

040

Henry Snow @henrys.now · 23/06/2026

I'm pretty critical of the Luddites but with critiques like these I see why we've gotten less-than-thoughtful defenses. The Luddites did not have a position on whether machines in general produced long term aggregate job growth, but whether they made their lives immediately worse

Ricardo and the Luddites alike.49 In a very different environment, it was widely feared that the mechanization of agriculture in Asia in the 1970s would lead to widespread rural unemployment; this did not occur (Campbell, 1990, p. 26). Recent studies by labor economists find that the introduction of new technology is on balance associated with positive job growth. One such study flatly declares that "job growth and the introduction of new technology appear to be complements rather than substitutes. The Luddites were wrong"

120

Henry Snow @henrys.now · 22/06/2026

one look at this and you can see why i write about these guys

An 1824 image of a shipwright. He's wearing slightly baggy clothes and squinting as he works with a hammer in one hand and a chisel in the other, standing on a dock beside a ship

151

Henry Snow @henrys.now · 19/06/2026

Yeah most left expression has long been (what we'd now call) cringe! Here's a subversive poem from 1802, for instance. It's calling for an illegal strike at a naval dockyard to demand food relief-- with rhymes about cheese. Very radical and very No Kings sign cringe

Farmers Bakers Butchers likewise
An example there is put before your eyes
It is the pride of your heart to see the poor starving
You that have got a plenty do not care a farthing
As for the horse likewise the Carter
It is high time they were taken the head shorter
In lowering 6d they would sooner rise two
They do it to try the poor what they will do
These Big dons will do just as they please
They would make the poor believe the moon is green cheese
If the yard will all be true and stand to these Rules
We will soon let them know who is the biggest fools

1947

Henry Snow @henrys.now · 18/06/2026

another highlight from this account: shipbuilders were marked absent in muster books with 3 dots, but present with a line in the same place, which meant they just bribed clerks to cross the former through with the latter in order to take personal days

Job work theoretically also meant pay was based on productivity, but artificers completely subverted it. Navy Board orders put a cap on daily earnings under job work--4s 2d. at Portsmouth, according to Mary—and in at least some cases this maximum within what was meant to be a productivity-encouraging system became the general rule. “By falsifications of various descriptions,” Mary explained, workers managed to bump their pay rates up to this level without increasing their productivity. One such falsification was creative use of extra. “The term Extra,” a special Commission of Naval Enquiry determined in 1805, used to refer only to what we would today call overtime but “since the introduction of task and Job work, it has been taken in a more extensive sense,” and now referred to “any mode of working, though within those hours, by which the wages of the men are increased beyond the established rate of daily pay.” Shipwrights were paid extra to work on Sundays, and further extra and provisions if they had to sleep aboard.
There was a “total want of system and regulation in the accounts of Job Work. . . the employment of the workmen by Job at Plymouth Yard, has been altogether fallacious.” Portsmouth, judging from Mary’s account, was no better. With some help from clerks and supervisors—quartermen, perhaps even principle officers— shipbuilders equalized wages.It was not as if anyone was going through the account books: the Clerk of the Cheque signed them, but, he conceded to the Commission, he was too busy with other duties to actually look at them before doing so. “The most laborious, the best skilled , the idle , the infirm , those employed on regular work at the dock side, or those buffeted about half their time in going out to Spithead,” Mary claimed, were all paid the same rate. Having neutralized task work with open rebellion in 1775, artificers took control of its successor system with quotidian subversion.

010

Henry Snow @henrys.now · 04/06/2026

doing some Learning

a, Abstract (top) and concrete (bottom) examples of a task inducing interference during learning. One stimulus input (seeing the water) triggers two prediction outputs (hearing the water and smelling the salmon). One output is correct (smelling the salmon), whereas the other output is an error (not hearing the water). b,c, Backpropagation produces interference during learning; not hearing the water reduces the expectation of smelling the salmon (b), although the salmon was indeed smelled. Prospective configuration, on the other hand, avoids such interference (c). In backpropagation, negative error propagates from the error output to hidden neurons (b; left). This causes a weakening of some connections, which, on the next trial, improves the incorrect output but also reduces the prediction of the correct output, thus introducing interference (b; middle and right). In prospective configuration, neural activity settles into a new configuration (different intensities of purple) before weight modification (c; left). This configuration corresponds to the activity that should be produced after learning, that is, is ‘prospective’. Hence, it foresees the positive error on the correct output and modifies the connections to improve the incorrect output while maintaining the correct output (c; middle and right).

100

Henry Snow @henrys.now · 27/05/2026

ominous stuff from @quinnslobodian.com and ben tarnoff in light of the literal pull the plug clause Musk apparently has in the Colossus compute deal with Anthropic-- a "kill switch" for if he decides they are "harming humanity," which for Musk could easily mean "Claude is insufficiently racist"

Musk does not just sell cars, rockets, or satellites. He sells the fantasy that, in an increasingly unstable world, both states and individuals can fortify their self-reliance by plugging into /his/ infrastructures. The paradox is that, in doing so, you become reliant on him . . . Trying to unplug from Musk, you realize he owns the socket.

0133

Henry Snow @henrys.now · 22/05/2026

comptroller of the navy hugh palliser does in fact look exactly like a guy who would start insane drama with his coworker

110

Henry Snow @henrys.now · 22/05/2026

"IF HYPOTHETICALLY a VERY GOOD MAN were boldly proposing my proposals and anyone should get in my, I mean his, way, that person would be a traitor and a scoundrel and I-- I mean anyone in my position-- would be a national hero" I told you all 18th-cent. timber supply debates could get heated

A bunch of text from Yeoman Lott's 1767 "Important Hints Toward and Amendment of the Royal Dockyards"; he's self-pityingly rambling

110

Henry Snow @henrys.now · 06/05/2026

writing a bit about that Florida knock-off right-wing AP curriculum and wow this is especially revolting

Slavery Uprising in Santo Domingo: Between 1791 and 1804, a slave revolt in Haiti resulted in a race war, cruelty, and bloodshed. Over 100,000 whites and 60,000 blacks were killed. The violence caused American statesmen to realize that an imprudent or rash approach to resolving slavery could result in a similar race war and bloodshed in
America.

0162

Henry Snow @henrys.now · 27/04/2026

incredible poster's energy from this person. i must identify them. i must

proposed making wool export from the colony punishable by death, while allowing tax breaks for widows on their hypothetical slavery tax. But True Son clearly intended the tax to destroy slavery as quickly as possible, and their rhetoric was uncompromising: they exhorted their fellow patriots to “exert yourselves in pursuing a stop to that hellish practice of deluding and enslaving another part of the human.” “How amazing is it!” a True Son asked, that America could cry out “very justly” against only “a distant prospect of slavery” under Britain and “[emphasis in original] at the same time be guilty (as they really are) of enslaving, and promoting the slavery, of thousands and tens of thousands of their fellow creatures!!!”

170

Henry Snow @henrys.now · 22/04/2026

he continues to haunt us all (antisocial egoist utopian, 80% right, 20% left in a politically irrelevant way that still shapes discourse)

0111

Henry Snow @henrys.now · 22/04/2026

NAM Rodger gets a lot wrong about 18th cent naval politics in my view but his narrative talent is undeniable, particularly when covering operations. "this he obligingly did" - excellent, understated, punchline

By this time it was too late. Byng’s squadron, now with thirteen ships of the line, met La Galissonière’s twelve and attacked, but it was handled with notable incompetence and only a few of the British ships, not including the flagship, were seriously engaged. The battle was a disappointment, but the British force was substantially intact and there was still everything to play for. The French position was so precarious that Byng needed only to cruise in the vicinity or let loose his frigates against the unescorted merchantmen bringing Richelieu’s supplies. The only hope for the French was for Byng to withdraw all British ships from the Western Mediterranean, and this he obligingly did. ‘What a scene Byng had open to him, and to throw it all away!’ exclaimed Boscawen when he heard.32 On 28 June, after an heroic resistance, Blakeney was forced to surrender.

110

Henry Snow @henrys.now · 21/04/2026

had to double check the notes to remind myself if i cut that bit (i did not) and wow this was a cursed series of footnotes

65. Beff Jezos (@BasedBeffJezos), Twitter, June 16, 2024, 7:28 p.m. ET.
66. “Pause Giant AI Experiments: An Open Letter,” Future of Life
Institute, March 22, 2023.
67. Keach Hagey, “The Secrets and Misdirection Behind Sam Altman’s
Firing from OpenAI,” Wall Street Journal, March 28, 2025.
68. Jeffrey Dastin et al., “OpenAI Appoints New Boss as Sam Altman
Joins Microsoft,” Reuters, November 21, 2023; Jeffrey Dastin and
Aditya Soni, “Sam Altman to Return as OpenAI CEO After His
Tumultuous Ouster,” Reuters, November 22, 2023; Luciana Lopez,
“Economist Larry Summers Joins the Board of OpenAI as Ousted
CEO Sam Altman Returns,” cnn.com, November 22, 2023.
69. Marc Andreessen, “The Techno- Optimist Manifesto,” Andreessen
Horowitz, October 16, 2023, a16z.com.
70. Thomas Sowell, A Conflict of Visions (William Morrow, 1987)

092