xuanalogue.bsky.social - Profile | ThreadSky | a Reddit-style client for Bluesky

I've been warning for years about Tech's interest in Nuclear and how the investment in Nuclear will not feasibly satisfy or meet AI's energy needs, and thus will ultimately lead to threats and compromise to Nuclear safety. And here we are. thebulletin.org/2025/02/trum...

submitted 1 hour ago • 1 comment

THEY'RE GOING TO ISSUE PERMANENT VISA BANS TO TRANS VISITORS TO THE US. This isn't just for athletes. The directive as stated applies to all visa applications made by trans folks and declare it material fraud to use a different gender marker on applications. www.theguardian.com/us-news/2025...

submitted 1 day ago • 247 comments

1. Major exclusive breaking news: Marco Rubio may have just banned trans foreigners seeking visas with correct gender markers from US Entry. A new cable, effective immediately, says visas must contain assigned sex at birth and should be rejected if they don't. Subscribe to support our journalism.

submitted 1 day ago • 131 comments

caught myself verbally saying "sksksksksk" just now the brainrot is coming for me

submitted 1 day ago • 0 comments

Was looking for a explainer on why there's so much more support for AfD in East Germany relative to West Germany, and this one was pretty helpful! newlinesmag.com/argument/the...

submitted 1 day ago • 2 comments

Computational-Statistical Tradeoffs at the Next-Token Prediction Barrier arxiv.org/abs/2502.12465 New paper (another fun internship project!) with Dhruv Rohatgi, Adam Block, Audrey Huang (ahahaudrey.bsky.social), and Akshay Krishnamurthy (akshaykr.bsky.social). 1/11

submitted 3 days ago • 1 comment

Why I 🧡 the web. Someone made an app where you LITERALLY HAVE TO TOUCH GRASS to unlock your apps (for #mentalhealth and laughs?) touchgrass.now #iOS

submitted 2 days ago • 5 comments

I love how tech keeps going back to humanoid robots even though literally any other design would be safer, more useful, and more stable for personal use. "no it has to be a SERVANT," screams the reptile hindbrain

submitted 2 days ago • 16 comments

yikes, isn't this almost surely illegal??

submitted 2 days ago • 1 comment

Free speech no longer exists in the US government. For @theatlantic.com I spoke with 12+ federal workers in 6 agencies who said the Trump administration’s actions have led to pervasive self-censorship, even on issues some view as critical to national security. 1/ www.theatlantic.com/technology/a...

submitted 3 days ago • 5 comments

Out today in Nature Machine Intelligence! From childhood on, people can create novel, playful, and creative goals. Models have yet to capture this ability. We propose a new way to represent goals and report a model that can generate human-like goals in a playful setting... 1/N

submitted 5 days ago • 5 comments

It was fun helping out on this paper! Haven't worked on nat lang ToM benchmarks in the past, and it was cool to see that SMC-like scaffolding of LLMs works better than o1-style reasoning models -- though both are still less reliable than explicit model-based ToM reasoning.

submitted 5 days ago • 2 comments

DEI initiatives are still legal. Universities need to stand their ground. A Friday gift to your university's General Counsel Office - courtesy of an all star lineup of civil rights lawyers and scholars. You're going to want to read this.

submitted 5 days ago • 13 comments

When they start also saying "sieg heil" while doing their Nazi salutes will it be reported as "made a gesture somewhat resembling a Nazi salute while making a noise somewhat resembling 'sieg heil'"

submitted 6 days ago • 4 comments

“The very fact that I am having to publish this anonymously in a country that has the right to free speech written into our Constitution is an indictment of what is happening” www.bmj.com/content/388/...

submitted 6 days ago • 0 comments

The move to use LLMs to enforce platform codes of conduct is going to expose all the intentional vagueness in the policies. LLMs can certainly enforce what is written in the policies, but I bet it’s going to yield a bunch of undesired changes to enforcement patterns.

submitted 6 days ago • 1 comment

DOGE By The Numbers theonion.com/doge-by...

submitted 6 days ago • 189 comments

New: DOGE's spending has been secret. No longer. My colleagues have uncovered it. www.propublica.org/article/doge...

submitted 6 days ago • 863 comments

🚨New Paper! So o3-mini and R1 seem to excel on math & coding. But how good are they on other domains where verifiable rewards are not easily available, such as theory of mind (ToM)? Do they show similar behavioral patterns? 🤔 What if I told you it's...interesting, like the below?🧵

submitted 6 days ago • 3 comments

An interesting piece from a perennially fun-but-grumpy writer (www.the-hinternet.com/p/my-kind-of...). He didn't like the "tumblr regime" of yesteryear but hates the "4chan regime" we have now even more. I think he's fundamentally correct that avant-garde internet norms have swept to power in waves

submitted 7 days ago • 7 comments

Upon learning that yesterday would be my last day as a program officer at the National Science Foundation, I shared this parting message with my colleagues. The next few months will be frenetic and stressful for them. Here are some things that you can do to help them with the mission ahead. (1)

submitted 7 days ago • 68 comments

Where are the women in AI? www.interface-eu.org/publications...

submitted 8 days ago • 0 comments

New investigation by @apnews.com on the IDF's use of OpenAI models in Gaza. I spoke to them regarding this first confirmation of commercial foundation models being directly used in warfare and the implications of using highly error-prone AI for life or death determinations apnews.com/article/isra...

submitted 8 days ago • 1 comment

submitted 8 days ago • 199 comments

Q: “How are cuts in science compatible with wanting to go to mars?” A: they believe that scientific progress primarily happens in industry. There are no problems remaining to solve, only “hardcore engineering”

submitted 11 days ago • 6 comments

The fact that only 2% of professors identify as fascists shows just how insular and out of touch universities have become.

submitted 13 days ago • 21 comments

aww thanks for promoting my content Google AI

submitted 10 days ago • 0 comments

On Monday, Feb. 17, 2025 (Presidents’ Day), coordinated protests will occur in all 50 states at or near each state’s capitol (and some city halls) to peacefully oppose the Trump Administration’s actions and “Project 2025” agenda. Find details for your state ⬇️🪧

submitted 10 days ago • 3 comments

BBC research into chatGPT’s accuracy answering questions about current events showed many issues with AI generated answers arstechnica.com/ai/2025/02/b...

submitted 11 days ago • 1 comment

Question for RL folks. In certain kinds of trajectory optimization, we allow the optimizer to consider trajectories that are physically impossible, but encourage it towards physical validity with a loss function (while also optimizing task loss). My Q: is there any analog in RL?

submitted 11 days ago • 3 comments

Doing the unthinkable: The deep cuts to the #CDC's workforce today are expected to decimate the Epidemic Intelligence Service, a program that has trained public health rapid responders for decades. Envy of the world. Poof! www.statnews.com/2025/02/14/t...

submitted 12 days ago • 545 comments

I've often struggled to explain _why_, even as a novice, I enjoy #julialang so much, often falling back to "It just feels nice." Apparently, I should have just let the Julia developers explain it. Came across the following from "Why We Created Julia" from 2012-02 and it all makes more sense now.

submitted 13 days ago • 2 comments

There hasn’t been nearly enough appreciation for this amazing paper by Clark Barrett and @rebeccasaxe.bsky.social Anthropologists have observed people in certain cultures blaming agents for behavior without regard to mental states (intent, knowledge, etc.). Why does this happen?

submitted 12 days ago • 2 comments

got my first hate-mail today lol T_T

submitted 15 days ago • 3 comments

One of the biggest areas of AI hype right now is the notion that it will hyperaccelerate scientific progress. I understand why people think this — AI is already accelerating scientific _production_. 🧵

submitted 17 days ago • 4 comments

Sri Krishnan temple wishes all Chinese devotees a happy lunar new year (along one of my favourite streets in Singapore)

submitted 17 days ago • 0 comments

@xtimv.bsky.social and I were just discussing this interesting comment in the DeepSeek paper introducing GRPO: a different way of setting up the KL loss. It's a little hard to reason about what this does to the objective. 1/

submitted 17 days ago • 3 comments