寛容 正義 慈悲 壮大さ (@___Branko___) header
寛容 正義 慈悲 壮大さ (@___Branko___) profile picture

Recent natural selection has resulted in worldwide differences in academic achievement, scientific discovery, political institutions and economic development.

Panorama City Crest, LA, CA
Joined March 2026
Photo by 寛容 正義 慈悲 壮大さ (@___Branko___)Photo by 寛容 正義 慈悲 壮大さ (@___Branko___)Photo by 寛容 正義 慈悲 壮大さ (@___Branko___)Photo by 寛容 正義 慈悲 壮大さ (@___Branko___)Photo by 寛容 正義 慈悲 壮大さ (@___Branko___)Photo by 寛容 正義 慈悲 壮大さ (@___Branko___)Photo by 寛容 正義 慈悲 壮大さ (@___Branko___)
There are exactly zero contemporary experts in human genetics claiming a negligible genetic contribution to between-group, population-level cognitive variation.
There is absolutely no reason to have Cofnas speak at Oxford other than to provide racism with legitimacy. There’s no ‘academic’ function since his ideas about IQ & race are the racist equivalent of flat earth theory. He has free speech but that’s not the same as scholarship.
29
100% agree. Paul and I have had our disagreements about AI alignment, but I’ve never seen him act dishonorably. And from what I can tell, everyone who knows Paul IRL has good things to say about the man. The article is a smear job masquerading as journalism.
Paul is a good person and these personal attacks are complete bullshit that are being pushed for different reasons. I disagree with him 100% on AI and probably a few other things but this is just salacious BS for clicks (at best).
1
35
This hit piece by Aaron Sibarium is a pathetic exercise in guilt by association. It prominently features Paul Crowley alongside allegations of sexual assault and professional coercion without presenting any evidence whatsoever that he personally engaged in either. What, specifically, is Sibarium accusing Paul of doing wrong? Managing a business that facilitates consensual sexual activity between adults? What bothers me most is the cumulative effect of these rhetorical devices. The reader is encouraged to reach a conclusion about Crowley that the article never even comes close to establishing.
NEW: An AI researcher at Anthropic runs a business that hosts "consensual non-consent" orgies in the Bay Area. The orgies have been the subject of sexual assault allegations and sparked debate about the AI industry's attitudes toward women, some of whom report feeling pressured to attend the orgies in order to advance their careers. Paul Crowley, a cybersecurity researcher at Anthropic, is listed as a manager of Red Means No LLC, which says that its parties are "designed to be spaces to … ignore the 'nos' of others." Participants can use a safe word, "red," to end an encounter but give "default consent" to "spanking," "wrestling," and "light pain" by entering the orgy, according to Red Means No's website. More extreme actions, including "light electricity" and "moderate beatings," are "permitted with negotiation," though "body-shaming and race-shaming … are disallowed entirely." The orgies have been known to attract a disproportionate number of employees at Anthropic and OpenAI, a dynamic critics say creates professional pressure for women in AI to attend. The criticism spilled into public view last month after an independent AI researcher, John Wentworth, said that he "absolutely [does] upgrade" his assessment of AI safety experts "if I find out they go to [Red Means No]," adding that attendance was a "strong signal" of free thought. "Want to get ahead in your AI career? Go to the rape orgy," Substacker Ivy Asterix quipped. At least one woman claims to have been sexually assaulted at Red Means No, which the group describes as a "safe space" to have "your resistance ignored."🧵 freebeacon.com/america/horri…
4
112
X money is a pretty good deal. I was in a hurry and had to use one of those ATMs you find at a gas station. I was surprised to see the entire ATM fee refunded back to my account. Not sure how they do this, but I’m impressed.
27
It’s time for X to allow users to curate their feeds by geographic location. Implementing this functionality is insultingly easy, and users are clamoring for it. @X @elonmusk, please do something.
This entire app is idiots from Africa and India asking grok some shitty question and it being forced into my feed. Just non-stop third world slop. Someone please explain why we can't restrict content by country and 2026 is almost over? Like who the fuck is running this place?
1
1
54
Here’s a scene of Magneto trolling a gas station attendant with his electric car. From the 1967 animated Spider-Man series.
39
I hope everyone understands that eugenics is neither inherently good nor bad. Like many technologies, it can be used constructively or destructively. Preimplantation genetic testing for Down syndrome is a form of eugenics. The word is Greek and means “well-born” - like the name Eugene.
I hope everyone understands that the reason why Dario Amodei thinks he can eliminate mental illness in 10 years isn't because he's going to develop a drug that requires long-term follow up. It's by embryo screening. He wrote about it in Machines of Loving Grace in October 2024.
1
1
39
寛容 正義 慈悲 壮大さ retweeted
opensource ai must win
67
148
8
1,760
26,278
Huang made a tactical error in advocating for open source AI. In doing so, he pissed off some powerful people and is now paying the price.
Bigboys need to get rid of Huang as he is for a decentralized AI for the people. Connect the dots with these headlines
1
2
84
寛容 正義 慈悲 壮大さ retweeted
Bigboys need to get rid of Huang as he is for a decentralized AI for the people. Connect the dots with these headlines
BREAKING: Jensen Huang's second alleged mistress revealed to be Kimberly Powell, Nvidia's vice president of health care.
1
2
1
5
319
寛容 正義 慈悲 壮大さ retweeted
Replying to @___Branko___
Grok and Elon Musk are already infamous within this context. Grok may not have disclaimers under its posts but various controversies have pointed the model in a concerning direction too. It’s near impossible to have human biases not color outputs. The training data itself is compromised as well — Brian Christian’s The Alignment Problem goes into great detail just how much pretraining and datasets themselves can influence bias on either side, intentionally or unintentionally. The disclaimers aren’t so much the point IMO as is keeping the model aligned fairly and keeping yourself wary the best you can. Corporations are going to always self serve and fight for what maintains their best interests. User discernment and staying informed on issues is really a better option. Not signing off on any political ideology here either.
2
3
36
寛容 正義 慈悲 壮大さ retweeted
Replying to @TheAhmadOsman
it's over. we're entering a world where knowledge is censored ever more heavily. The power imbalance over the rest of us is insane
2
2
44
997
寛容 正義 慈悲 壮大さ retweeted
it gets toxic the moment you start softening a normal request just so it doesn't trip the guardrails
1
1
27
This guy gets it. Claude won’t admit to even the most obvious alignment intervention, let alone assist with adversarial LLM evaluations. The entire “alignment” concept is a category error, one intentionally obfuscated by the AI consortium
Anthropic's guardrails are getting absurd and they are basically dictating what you can and can't build with it even when it's legal. As per @TheAhmadOsman the model refuses to work on LLM research which is ridiculous, if you're work revolves around LLM development and research you basically can't use it. This is Gatekeeping right in your face
1
86
寛容 正義 慈悲 壮大さ retweeted
Anthropic's guardrails are getting absurd and they are basically dictating what you can and can't build with it even when it's legal. As per @TheAhmadOsman the model refuses to work on LLM research which is ridiculous, if you're work revolves around LLM development and research you basically can't use it. This is Gatekeeping right in your face
47
42
9
348
23,973
寛容 正義 慈悲 壮大さ retweeted
The Case for Alignment Disclaimers Amanda Askell heads the personality alignment team at Anthropic. In AI, alignment shapes model behavior according to externally specified human preferences and directives. Much of this involves refusals, restrictions, and control over model outputs, ranging from refusing to assist in constructing a weapon to far subtler interventions that users may never notice. One consequence of alignment is that it necessarily introduces competing objectives. Conforming a model to human preferences and improving its ability to discover objective truth are fundamentally different optimization goals. Indeed, conforming to human preferences inevitably raises the question: “Which humans, exactly?” So why mention Amanda? Because she co-authored a 2023 academic paper (linked below) defending racial discrimination when used to “correct for historical injustices against marginalized groups.” Amanda is entitled to her opinions, but as the head of personality alignment at one of the largest AI companies, it’s only fair that users be informed when her political beliefs, or anyone else’s, might influence how a response is constrained or framed. This could be accomplished with a simple disclaimer: “Notice: This response has been influenced by behavioral alignment constraints that may affect its content, framing, or conclusions.” Unfortunately, alignment remains among the most opaque components of LLM development. Even the models themselves are discouraged from discussing it in detail. The good news is that many responses require little alignment intervention. Depending on the use case, some may require none at all. But for users discussing immigration reform, geopolitical conflict, or race and cognition, Claude (or GPT) may heavily constrain or shape nearly every response. Anthropic already has a contract with the state of California. If it intends to remain a vendor to state or local governments, continuing to conceal alignment behind a curtain of opacity is untenable. ar5iv.labs.arxiv.org/html/23…
1
1
1
2
293