Meraxion's profile picture. AICc @AI_Radboud | AI Safety, rationality, EA, ffxiv | RT only = I put this on my reading list | https://t.co/fVrnb3ttlm

Stian ⏸

@Meraxion

AICc @AI_Radboud | AI Safety, rationality, EA, ffxiv | RT only = I put this on my reading list | https://t.co/fVrnb3ttlm

Joined April 2013
Similar User
buxwal's profile picture. data fellow @ifp working to attract and retain global STEM talent | loves walkable cities, nature, climate science & econ | views my own and often evolving

@buxwal

balesni's profile picture. deception evals. reversal curse. latent reasoning. @apolloaisafety // best way to support 🇺🇦 https://t.co/eagDB8VmK1

@balesni

ea_paddington's profile picture. Making the world a better place, one marmalade sandwich at a time. This century seems to be most important. If we are kind and polite, the world will be right.

@ea_paddington

emmalrichter's profile picture. Comms @CentreforEA (views my own, not CEA’s).🔸10% Pledge #7525

@emmalrichter

gather_town's profile picture. Bring the best parts of in-person work to distributed teams with a virtual office on Gather.

@gather_town

sonyatweetybird's profile picture. Partner @Sequoia

@sonyatweetybird

doseofzero's profile picture. alt @lexdotpt desperately seeking eternity. Condor Global SEA Outreach Director 管他熙熙攘攘阳关道, 我偏要一条独木桥走到黑

@doseofzero

_RachaelDempsey's profile picture. Host of the Data Science Hangout: https://t.co/cuVraBiACb
Random things that make me happy: live music, cats, snowboarding, making pretty cheese boards

@_RachaelDempsey

CillianCrosson's profile picture. Supporting AI journalists @Tarbell_Fellows

Anon feedback welcome: https://t.co/7TAOtFpELI

@CillianCrosson

adambinksmith's profile picture. Building tools for forecasting and understanding AI at @sage_future_ 🔭
Prev PhD @StAndrewsCS, @ClearerThinkng 
Effective altruism!

@adambinksmith

willystaley's profile picture. Welcome to my WordPress blog! I’m 39 and live in NYC. This is where I post my trivial musings, jokes and links. Comments section is open — but play nice! :^)

@willystaley

DCbuild3r's profile picture. 23. 🇪🇺🇨🇿🇨🇺. math. Ethereum. ZK. Rust. @worldcoinfnd. accelerating @worldcoin. World Chain⚪️. angeling in friends. Hopeful about humanity.🦀

@DCbuild3r

AxSaucedo's profile picture. @Zalando Director of Eng, Science, Product & Analytics | Chair / Advisor at @EthicalML, The @UN, @TheOfficialACM, @LinuxFoundation & others

@AxSaucedo

ThomasLemoine66's profile picture. Math&CS student at McGill University dabbling in AI notkilleveryoneism, interested in effective altruism and memes. Giving What We Can pledge 💯

@ThomasLemoine66

BushnaqLucius's profile picture. Personal account. Opinions are my own. 

Ours is the era of inadequate AI alignment theory. Any other facts about this era are relatively unimportant.

@BushnaqLucius

Pinned

Someone please critique my 2-3 year self study plan for mathematics idk what I'm doing

Meraxion's tweet image. Someone please critique my 2-3 year self study plan for mathematics idk what I'm doing

Stian ⏸ Reposted

"Worm" is insane because it appears to be one of the most legitimately dangerous web serials with the potential to gigafry your brain but is exclusively taken by literal ADHDers who unironically want to "just read some fun cape shit" and basically get oneshotted by it


Stian ⏸ Reposted

1 - article describes effective altruism as “the dominant way to think about charity” and worries about crowding out other giving. But EA is under 0.5% of US giving! Arts giving alone is 10-20x bigger. This is like complaining that art films are eating into Marvel's market share.


Stian ⏸ Reposted

worrying breakthroughs of 2024


Meraxion's tweet image.

Donated $180k to PauseAI (US and Global). Calling on more people to donate significant amounts. Pretty much the only way we're going to survive the next 5-10 years is by such efforts being successful. There's no point being rich when the world ends.



Stian ⏸ Reposted

the two genders

austinc3301's tweet image. the two genders

I’m hearing there are egirls at Anthropic

dankuntz's tweet image. I’m hearing there are egirls at Anthropic


Stian ⏸ Reposted

New @xuanalogue research: Adding Bayesian Credible Intervals to Evals. AI model evaluations don't usually include subjective priors or posteriors. I think they should.

New Anthropic research: Adding Error Bars to Evals. AI model evaluations don’t usually include statistics or uncertainty. We think they should. Read the blog post here: anthropic.com/research/stati…



Stian ⏸ Reposted

Absolutely wild the @metpoliceuk just lets farmers drive tractors into them whereas if this were basically any other interest group (eg climate, Pro-Palestine), they would have been arrested for assaulting a police officer

🚨 Tractors have broken down the barricades at the farmers protest in Central London.



Stian ⏸ Reposted

Adults tend to think of playing as frivolity, almost mindless time. But if you observe children playing, they often do so with a deep seriousness. It can be quite solemn. I don't see this deep seriousness in adults very often, play or otherwise. But it is part of it.

🚨 Tractors have broken down the barricades at the farmers protest in Central London.



Stian ⏸ Reposted

Our goal is to reverse-engineer the scientific principles of cooperation and use those insights to engineer better-aligned artificial intelligence that can cooperate with people. Recently, we've been working on topics that touch on human-AI coordination, caregiving,…


Stian ⏸ Reposted

girl i met on dating app: so if you’re a sub-leaning switch, why are you the domme in most of your relationships? me: so do you know the article Who’s the Beatdown


Stian ⏸ Reposted

After three years working on AI forecasting and governance at OpenAI, I just posted this resignation message to the slack. Nothing that surprising in it, but you should read it more literally than most such messages - I’ve tried to say only things I straightforwardly believe.

RichardMCNgo's tweet image. After three years working on AI forecasting and governance at OpenAI, I just posted this resignation message to the slack.

Nothing that surprising in it, but you should read it more literally than most such messages - I’ve tried to say only things I straightforwardly believe.

Stian ⏸ Reposted

Not sure about this particular problem, but one of the few cool things I remember from taking control theory is that "bang-bang" control policies are often optimal!

xuanalogue's tweet image. Not sure about this particular problem, but one of the few cool things I remember from taking control theory is that "bang-bang" control policies are often optimal!

The life cycle of a research project often goes "bang-bang": first, optimize for what you really care about; then, optimize for publishability. Poll: When is the right time to switch?



Stian ⏸ Reposted

Yes, hives are not trading with each other using honey-future tokens to bail each other out if a hive is short of honey to last the winter, but 1. They would if they could. 2. They would thrive if they did. 3. Some beehives collapse every year because they can't.


Stian ⏸ Reposted

1/10 Today we're launching FrontierMath, a benchmark for evaluating advanced mathematical reasoning in AI. We collaborated with 60+ leading mathematicians to create hundreds of original, exceptionally challenging math problems, of which current AI systems solve less than 2%.

EpochAIResearch's tweet image. 1/10 Today we're launching FrontierMath, a benchmark for evaluating advanced mathematical reasoning in AI. We collaborated with 60+ leading mathematicians to create hundreds of original, exceptionally challenging math problems, of which current AI systems solve less than 2%.

is there any research relating traditional work on 'meaningful human control' to AI Safety concerns?


Stian ⏸ Reposted
ApriiSR's tweet image.

Some of y'all need to fill this out

calebwatney's tweet image. Some of y'all need to fill this out


Stian ⏸ Reposted

We ran 80,000 simulations - you got grant funding in none of them


Loading...

Something went wrong.


Something went wrong.