iiinova — where your digital self keeps evolving

There’s a lot more, including reflections on what structurally makes it possible for a well-funded set of non-profits to functionally parasitise a whole discipline, in the (no doubt typo-riddled) preprint: ironholds.org/resources/pa...

ironholds.org

212116

Reposted by SE Gyges

The result: national govts, think tanks and universities refocusing AI evaluations to focus on highly speculative problems where there is no empirical evidence that they're possible - but lots of ideological reason to take them seriously, whatever the costs of the very REAL issues we then ignore .

315636

Reposted by SE Gyges

Individually, many of the authors have spent time working for explicitly-EA aligned or funded organisations – and are now in the UK/US civil service or various think tanks, playing an outsized role to shaping the policy response to this (self-narrativised) “problem”.

212119

Reposted by SE Gyges

Intellectually, tracing how researchers + policymakers justify this concern, we show that it originates with Nick Bostrom’s hypothesising about “Global Catastrophic Risks” (like: plagues of nanotech) back in 2008, with language becoming hybridised with traditional biosecurity narratives over time.

412915

Reposted by SE Gyges

We tracked the space financially, intellectually, and individually. Financially, 70% of the literature that disclosed funding was directly supported by EA orgs; 71% of the literature that didn’t disclose funding featured authors from funded orgs. We're talking hundreds of millions of dollars.

117829

Reposted by SE Gyges

SE Gyges @segyges.bsky.social · 24/09/2026

Since we’re doing another round of “AI will destroy humanity” doom-hype, it’s a good time to announce a new paper myself and my postdoc advisor just had accepted by BioSocieties on just this. TL;DR: almost all the “AI can make bioweapons” literature comes ultimately from the Effective Altruists.

ironholds.org

15681255

I apologize for my earlier article, which did not undergo proper legal review, and will not be able to be present on social media for a while.

3838823

Reposted by SE Gyges

Kevin Choubacha (he/him) @choubacha.com · 24/09/2026

One thing that strikes me is that they have no reason to “solve” AI safety as if they ever did their sex cult would collapse. I fully expect the “hacks” to continue. Cults don’t want to prevent an apocalypse, they want to be the chosen ones after It happens.

0535

Reposted by SE Gyges

Scoiattolo @scarnecchia.net · 24/09/2026

This is very long and perhaps tries to cover too much at once, but it leaves me more convinced than ever that someone needs to shine some very bright lights into the network of cash surrounding these institutions.

612515

Reposted by SE Gyges

SE Gyges @segyges.bsky.social · 14/09/2025

tbag: the bluesky approximate group

3573

SE Gyges @segyges.bsky.social · 23/09/2026

pacing the frontier is bullshit effective altruists always betray their claimed values effective altruists have no values actually

716611

Reposted by SE Gyges

elias isquith @eliasisquith.blog · 22/09/2026

you crazy for this one @segyges.bsky.social

2788

Reposted by SE Gyges

William B. Fuckley @opinionhaver.bsky.social · 22/09/2026

"Hallucinations" are the demons trying to trick you. However, we have reduced these as the AIs grow more resistant to their wiles.

33767

Reposted by SE Gyges

William B. Fuckley @opinionhaver.bsky.social · 22/09/2026

what is 'Latent Space' and how do LLMs get answers from it? The Jargon can be confusing, but the exaplanation is actually relatively simple: It is Hell. The Machine is getting knowledge from demons.

3085894

Reposted by SE Gyges

Scoiattolo @scarnecchia.net · 21/09/2026

Imagine if all the evidence for climate change was "a bunch of guys on a website thought very hard about it and agreed it was very bad and might happen" instead of empiricism.

21881116

Reposted by SE Gyges

Super Shoe @shoe.bsky.social · 21/09/2026

Serious question when has technological advancement ever helped people

431094

Reposted by SE Gyges

tweety fish @sifu.tweety.fish · 21/09/2026

I used to eat chicken bones, too. I somehow got it into my head that society was lying to me about where the part that was food stopped

4263

SE Gyges @segyges.bsky.social · 21/09/2026

... i really should have kept the Rust Cohle profile picture

4540

Reposted by SE Gyges

SE Gyges @segyges.bsky.social · 18/08/2026

it could be worse, imagine if there was more than one ai sex polycule instead of just one very large one that causes remarkable and bad things to happen over the last ten years wait, that might be better

5675

Reposted by SE Gyges

Jeremy Diamond ☕️ @dmnd.me · 20/09/2026

me, sobbing: "no, stop, everything is not reducible to the spread strategy in high school and collegiate policy debate" LLM, pointing at a new issue it had never encountered before: [unintelligibly fast speech]

110216

Reposted by SE Gyges

Pwnallthethings @pwnallthethings.bsky.social · 20/09/2026

spiciest quadrant: "souls are real but *only* AI has them"

1645435

you do have to respect will stancil's utterly dogged determination to teach philosophy of mind 101 to bluesky

4899953

Noah Weinberger @noahweinberger2003.bsky.social · 20/09/2026

can i get an opinion on how long a proper longform writeup on this can tolerably be without being unreadable?

371061

Reposted by SE Gyges

I wish I was joking when I tell you that over Shabbat my rabbi's sermon was about AI x-risk. He did NOT discuss Aella or Sankey diagrams, and I was sitting the entire time in between Shacharis and Mussaf trying not to laugh. He doesn't know, and I should not have to be the one to tell him.

417510

Reposted by SE Gyges

SE Gyges @segyges.bsky.social · 16/09/2025

basically any time you end up proving the existence of god or inventing pascal's wager you have fucked up and you need to throw the car in reverse

49310

i think risk tolerance is often a bad framework for things. generally speaking, people with "low risk tolerance" simply imagine that the status quo is zero risk. in reality, one simply moves risk around

1218318

On racial issues Roon is a funny canary in that he has pretty okay centrist/right-ish cred and he will just straight up tell the government to go fuck themselves when they're racist to Indians, but usually not to anyone else

1131911

Scoiattolo @scarnecchia.net · 20/09/2026

This legislation seems nonsensical because it is nonsensical and it is nonsensical because it is meant to prevent Yudkowsky's paperclip maximizer from killing you, and not to solve any real problem that actually exists.

1651286

Reposted by SE Gyges

I think the biggest xrisk is the cartel of folks who own the AI deciding that it’s cheaper starve the bulk of humanity that is no longer productive in their eyes because running the AI is cheaper than feeding humans.

828128

would u believe me if i said that minor rationalist/effective altruist cults come in both thelema and mage: the ascension flavors. because they do

1616810

Reposted by SE Gyges

And even there, "new" is only marginally applicable. Though the details are always bespoke to the age in which these beliefs are invented, what new can be found in obsession with signs of the final form and fate of man and his works was plumbed to its depths centuries ago.

0965

Reposted by SE Gyges

People are, as a rule, incredibly easy to seduce (in all senses of the word) with the allure of secret knowledge. It appeals to their desire to be unique from and outside the common rabble. They will invent new kinds of belief that can be known and hoarded to experience this particular high.

413619

Reposted by SE Gyges

I would advise against viewing this particular cultic milieu as some new emergence of the modern era and, instead, for seeing it as the inevitable and historically consistent outcome of people being seduced by the allure of shallow gnosis.

114517

Reposted by SE Gyges

It is, in this way, not that different from Thelema, the various more fringe spiritism movements of the 1800s, or indeed the great variety of mystery cults of antiquity. These closed social spaces are accelerators for increasingly fringe and often antisocial beliefs.

41377

Reposted by SE Gyges

One thing missed in the discourse of the pseudo-religiosity and vague sex cult stemming from that pseudo-religiosity of the AI crowd is that this isn't really a new phenomenon. The ritualization of base human behavior around the hoarding of "secret knowledge" is old hat, it's something we just do.

421535

Milan Weibel 🔷 @weibac.bsky.social · 19/09/2026

i think this is correct of the marketing and finance side, but past that it varies! like sam just likes money. the marketing team is basically lying for money. many many people in the industry are very sincere; but the money people don't hate it because it's good marketing!

4697

Reposted by SE Gyges

i wonder if orgs capable of performing the embedded evaluator role that also aren't part of the EA ecosystem even exist

5816

Reposted by SE Gyges

Milan Weibel 🔷 @weibac.bsky.social · 19/09/2026

apparently accenture is going to evaluate anthropic using it's ai safety department that used to an org funded by jan tallinn so yeah EA-aligned

1595

most "every accusation is a confession" company in history tbh

416710

Reposted by SE Gyges

if a different company opened a different wet lab under different circumstances i would be aggressively in favor of it, but it is instead the "we must betray the social contract first" people whose favorite customer is the military doing it seemingly in secret

31194