jimrandomh's profile picture. Long-time rationalist, currently working full-time on the software of LessWrong.

Jim Babcock

@jimrandomh

Long-time rationalist, currently working full-time on the software of LessWrong.

고정된 트윗

This doesn't quite fit in a Twitter-bio flag emoji. Here's what I believe:

jimrandomh's tweet image. This doesn't quite fit in a Twitter-bio flag emoji. Here's what I believe:

X used to occasionally switching you from the Following feed to the For You feed. Presumably, this was to get you to try it. It now switches on 100% of refreshes. This crosses a line, for me. First, because it signals an intention to frog-boil towards propaganda. Second, because…


Jim Babcock 님이 재게시함

I like to call it the “definite article fallacy” My friends and family love me for this btw


The clan that responds to every death by gathering everyone to inspect the body knows if it is under covert attack. The clan that fails to do this does not always know. Applying evopsych to human-specific adult instict is hard because our ancestors could think too, and if…

My mom died recently and I was surprised to find I had a bunch of very strong impulses that made no sense. I think there's some weird ancient shit baked in there



When people analyze world events, sometimes they apply an action-movie bias. Alt interpretation: If your diplomatic meeting is especially tedious, Xi Xinping can't be arsed to attend, so he sends a body double. Then his staff removes the body double's DNA, so you can't prove it.

Xi Jinping has his DNA scrubbed from surfaces for fear of bioengineered weapons

rSanti97's tweet image. Xi Jinping has his DNA scrubbed from surfaces for fear of bioengineered weapons
rSanti97's tweet image. Xi Jinping has his DNA scrubbed from surfaces for fear of bioengineered weapons


Jim Babcock 님이 재게시함

For the first time in six years, MIRI is running a fundraiser. Our target is $6M. Please consider supporting our efforts to alert the world—and identify solutions—to the danger of artificial superintelligence. SFF will match the first $1.6M! ⬇️

MIRIBerkeley's tweet image. For the first time in six years, MIRI is running a fundraiser. Our target is $6M.

Please consider supporting our efforts to alert the world—and identify solutions—to the danger of artificial superintelligence.

SFF will match the first $1.6M!

⬇️

If you are responsible for any web apps using React 19, you need to immediately drop everything and patch for CVE-2025-55182. You are in a race against automated security scanners.


When I use LLM coding tools like Cursor Agent, it sees my username in code comments, in paths like /home/myusername/project/..., and maybe also explicitly in tool-provided prompts. A fun experiment to run, that I haven't seen yet: If instead of my real username it saw a…


X now has an "about this account" page which says which country it thinks the account is from. If you're going to use it, use it quickly, because everyone who wants to lie about their country of origin will be working on ways to defeat it.


The important thing to understand, when you hear about things like this, is that most hacking groups do not have tradecraft nearly this shoddy, so the attacks you do hear about are vastly outnumbered by the ones you don't. Cyberespionage has been offense-dominant for decades.

We disrupted a highly sophisticated AI-led espionage campaign. The attack targeted large tech companies, financial institutions, chemical manufacturing companies, and government agencies. We assess with high confidence that the threat actor was a Chinese state-sponsored group.



Phone and computer OSes sometimes bundle theme/styling changes in with their updates, eg Liquid Glass. Suppose themes were fully separate from security and compatibility updates, and you had to precommit to either always accept, or always reject these updates. Would you rather:


It's fine. That beam is solid steel. Even if it's not steel, it's more than thick enough. Even if it's not thick enough, those workers have hard hats. And if hard hats weren't sufficient, OSHA would put a stop to it. And if OSHA was shut down, falling is impossible because…


"Confidence high because we are hiding."

You can tell an AI model that you won’t read their “hidden scratchpad" but they won’t necessarily believe you

JeffLadish's tweet image. You can tell an AI model that you won’t read their “hidden scratchpad" but they won’t necessarily believe you


Jim Babcock 님이 재게시함

"If Anyone Builds It, Everyone Dies" is now out. Read it today if you want to see with fresh eyes what's truly there, before others try to prime your brain to see something else instead!

ESYudkowsky's tweet image. "If Anyone Builds It, Everyone Dies" is now out.  Read it today if you want to see with fresh eyes what's truly there, before others try to prime your brain to see something else instead!

Jim Babcock 님이 재게시함

In the last few years we've seen: - The plot to kidnap Gretchen Whitmer - The Storming of the Capitol and pipe bombs left at the RNC and DNC - The break-in to kidnap Nancy Pelosi and the brutal on Paul Pelosi - Multiple assassination attempts against Trump - The assassination…


Want people to know that your writing wasn't produced by an LLM? Put some em—dashes in it, but use them incorrectly. No one will delve into that accusation. It's not just a signal, it's a _costly_ signal.


Loading...

Something went wrong.


Something went wrong.