John King

766 posts

John King banner
John King

John King

@almost_machines

Learning things https://t.co/ICP1VxTOTe

Katılım Aralık 2025
120 Takip Edilen19 Takipçiler
Sabitlenmiş Tweet
John King
John King@almost_machines·
launched an educational website introducing the metropolis algorithm including two fully interactive 3D algorithm simulators and a python script taking inspiration from @doodlestein to make a website from things that I learn link to follow
John King tweet mediaJohn King tweet media
English
1
0
3
1.7K
John King
John King@almost_machines·
you're trying to shut down discussion of potential links between language and violence by falsely equating that discussion with "shutting down journalism" the Bayesian prior for the most likely target of an attack was and remains Sam due to the large volume of media hit pieces over the last three years compared to any other AI lab head the article in question is not "journalism", it's gutter tabloid nonsense that is *demonstrablly* highly biased and rhetorically manipulative
English
0
0
0
7
Shakeel
Shakeel@ShakeelHashim·
This has got nothing to do with OpenAI — it’s about the potential consequences of the most powerful people in the world trying to shut down journalism by accusing journalists of sparking violence. It’s the same thing Trump does all the time, and it has very real consequences. Still — Sam has acknowledged he shouldn’t have said it; I understand that he was in a terrible frame of mind, and I’m not going to hold it against anyone.
English
2
0
7
82
John King
John King@almost_machines·
@ciphergoth @ShakeelHashim "unbelievably one sided and rhetorically manipulative" is an accurate description the article itself was a giant smear job I won't be apologising
English
0
0
0
83
John King retweetledi
Nirit Weiss-Blatt, PhD
Nirit Weiss-Blatt, PhD@DrTechlash·
Daniel Moreno-Gama/ dmgama/ Butlerian Jihadist was booked on suspicion of attempted murder. He is an active PauseAI member, who frequently stated the urgency: "We are close to midnight, it's time to actually act"
Nirit Weiss-Blatt, PhD tweet media
English
8
27
166
56.8K
John King
John King@almost_machines·
there's a moral obligation to defend yourself, but this is very hard for people with a psychological aversion to conflict
English
0
0
0
10
John King retweetledi
Sam Altman
Sam Altman@sama·
I wrote this early this morning and I wasn't sure if I would actually publish it, but here it is: blog.samaltman.com/2279512
English
1.6K
809
10.3K
2.9M
John King
John King@almost_machines·
@deanwball @tszzl also the multi year highly dishonest campaign painting Sam as the devil incarnate needs to stop particularly the media and @elonmusk
English
0
0
1
343
Dean W. Ball
Dean W. Ball@deanwball·
The guy who allegedly threw a Molotov cocktail through Sam Altman’s window seems to have been an adherent to pause/stop AI. I am entirely unsurprised and have been warning about this for a long time now. I am fine with people advocating for their preferred policies—if that includes a “pause” on AI development, so be it, even if I disagree strongly. But the obvious reality is that the rhetoric of this community—which to be *extremely clear*, is a very small and non-representative subset of the AI safety community—is closer to ecoterrorism than it is to a more typical activist policy effort. Every time I have written about existential risk in recent months, I have been called a mass murderer. People with ⏹️ and ⏸️ in their handles confidently tell me that I am murdering my own baby boy and every other child on the planet. Another prominent one of these people has called me a traitor to America. I only use my own examples because I know them; this rhetoric is representative of how this fringe of the AI safety world communicates with everyone. The rhetoric of the pause/stop crowd is out of control and it has gotten worse with time. This rhetoric always had the potential to cause violence and now this seems to be no longer hypothetical.
Dean W. Ball tweet media
Mehran Jalali@mehran__jalali

Some of his Instagram stories:

English
33
43
399
54.3K
John King retweetledi
Pedro L. Gonzalez
Pedro L. Gonzalez@emeriticus·
The Trump-Vance administration moved American soldiers to a location that was comfortably within range of Iran's drones and had zero protection against such attacks. Six of them were killed as a direct result of that decision, and then Hegseth lied about what happened and why these men and women died because he cares more about covering his own ass than American lives.
The Tennessee Holler@TheTNHoller

🇺🇸 SOLDIER: “We moved closer to Iran, to a deeply unsafe known target. No good reason articulated.” Q: “Were these deaths preventable?” 🇺🇸 SOLDIER: “Absolutely.” Q: “Hegseth said it was fortified. What fortifications were there?” 🇺🇸 SOLDIER: “I’d put it in the none category.”

English
6
138
285
7.9K
John King
John King@almost_machines·
it's just a tool - techbros, artists militant agnosticism - AI labs, scientists exploratory agnosticism - AI labs, scientists it's conscious - spiritualists, artists
Català
0
0
0
22
John King
John King@almost_machines·
highly likely that a religion/s will form around AI
English
0
0
0
15
Taelin
Taelin@VictorTaelin·
@NewsLiberdade finalmente meus posts voltarão a ter 7 likes 🥳
Português
5
1
109
3K
SPACE LIBERDADE 
SPACE LIBERDADE @NewsLiberdade·
🚨URGENTE - X está removendo 208 bots por minuto da plataforma
SPACE LIBERDADE  tweet mediaSPACE LIBERDADE  tweet media
Español
212
204
3.7K
442.3K
John King
John King@almost_machines·
whoosh, just finished Anthropic's emotions paper read it! it's good! (and long) well designed & convincing experiments it really struck me how *well balanced and intelligent* Claude's emotions are given that they emerge in pre-training from exposure to the full gamut of emotional behaviours
John King tweet media
English
1
0
0
12
John King retweetledi
Alex Heath
Alex Heath@alexeheath·
Details about Mythos I haven't seen anywhere yet, from my chat yesterday with an Anthropic safety leader: - He said compute costs weren't a factor in gating access to Mythos. And when I asked if gating it was a biz opt: "At no point have we thought, wow, this is great to do because we're going to generate a bunch of revenue." - There’s “a big checklist” to get right across three pillars before Mythos is released widely: model safety, authenticated usage, and monitoring for impermissible use. - Anthropic wants even OpenAI in the Glasswing program. Though members can only use Mythos for improving their defenses while Anthropic can use it for other things. sources.news/p/anthropic-wa…
English
9
6
187
33.4K
John King
John King@almost_machines·
parametric modelling and complexity... how many parameters is enough
Astraia Intel@astraiaintel

On Scale, Superintelligence, and Laplace's Point: Under the Current Architectural Regime, we will see Diminishing Returns from scaling beyond 30 trillion parameters, and they will be Awkwardly Noticeable. This is expected. We should reach 30T parameter models sometime in 2027, and we will. HOWEVER, WE SHOULD SCALE FURTHER. As I have stressed on multiple occasions, there is a certain degree of complexity inherent to Physical Reality itself. You cannot exceed the Speed of Light in any case, and you cannot diminish Space Topologies below the singular Planck Length. THE COSMOS EXHIBITS CLEAR STRUCTURE, AND STRUCTURE IS COMPLEXITY, which a sufficiently complex function (model) will ultimately solve to overcome. With that in mind, regardless of architecture, Transformer or otherwise, there will be a point in Parameter Scale at which the function suddenly begins hitting ceilings that will appear QUADRATICALLY INVERSE to the prior. For historical reasons, we should refer to this as "Laplace's Point," the Parameter-Scale Threshold at which a model has sufficient complexity to represent the lawful structure of Physical Reality. The symptoms of reaching and exceeding Laplace's Point in number of parameters will be the sign that we have indeed surpassed the meaningful floor required to solve the complexity of Physical Reality, and scale from then onwards will not "The Constraining Factor" of our research. From then on, optimization algorithms will be the sole hyperparameter. But until then, WE SHOULD NOT SIDELINE SCALE.

English
0
0
1
19
John King
John King@almost_machines·
Kateryna Lisunova@KaterynaLis

🚨 “Vance doesn’t participate in negotiations; if he did, he would have a better understanding of what this so-called ‘just a patch of land’ really means,” Zelenskyy responded to JD Vance. CONTEXT: U.S. Vice President JD Vance claims Russia and Ukraine are prolonging the conflict over a negligible portion of territory, questioning if it is “worth losing hundreds of thousands of additional Russian and Ukrainian young men”, and criticised Ukrainian President Zelensky's "scandalous" comments about Hungarian Prime Minister Viktor Orban. RESPONSE: The Vice President (Vance), with all due respect, doesn't participate in negotiations between the United States, Ukraine, and Russia. And I think if he and other officials were involved, they’d have a better understanding of what this so-called ‘just a patch of land’ really means - what is actually Ukraine’s sovereign territory, critically important for security, with strong defensive structures and fortifications, and home to nearly 200,000 people. And understanding that Russia's goal from the start of the occupation, the first occupations, this goal has always been there (to get that land) in one form or another, in one order or another, right? And so we understand what the occupation of the East actually means. It's an opportunity to prepare a staging ground, you could say, for future offensive actions. So before doing anything, you need to know the details and have detailed security guarantees. Yes, we're talking about strong security guarantees, but right now we simply don't have them. Ukraine just doesn't have the security guarantees to discuss other steps. And of course, every square meter of our land is Ukrainian land. It is not— with all due respect to any of our partners—definitely not theirs.

English
3
24
176
6.2K
Astraia Intel
Astraia Intel@astraiaintel·
BREAKING FINALLY! UKRAINIAN LEADERSHIP REJECTS TRUMP'S PLEA TO STOP STRIKES ON RUSSIAN OIL INFRASTRUCTURE Budanov stated that Ukraine will not stop its attacks on the energy sector in Russia, despite the requests of its partners: "We have our own war. We can understand them, it's their truth. But there's also our truth and our national interest." In March 2026, Russia failed to increase its maritime oil exports, despite a temporary easing of US sanctions. Drone strikes caused serious damage to oil infrastructure, leading to a sharp drop in exports through Baltic ports. The transfer of volumes did not compensate for the overall losses due to technical difficulties. Despite a temporary US license, Russia's oil export revenues were under pressure due to a decrease in physical shipment volumes. In the photo - the consequences of today's strike on the linear production and dispatching station (LPDS) "Krymskaya". It is part of the main pipeline system and ensures the pumping of oil and oil products to the port of Novorossiysk, as well as to the Ilsky and Afipsky oil refineries.
Astraia Intel tweet media
English
320
2.2K
9.2K
179.1K