JavaScript is not available.

Zack Witten

@zswitten

Dec 7, 2022

Discovering Latent Knowledge in Language Models Without Supervision is blowing my mind right now. Basic idea is so simple yet brilliant: Find a direction in activation space where mutually exclusive pairs of statements are anticorrelated. I <3 clickbait so: the Truth Vector.

8

107

759

Alexander Berger

@albrgr

Nov 24, 2022

🚨🚨🚨 GiveWell's end of year recs are out!🚨🚨🚨blog.givewell.org/2022/11/23/giv

2

23

97

Dan Hendrycks

@DanHendrycks

Oct 11, 2022

Can ML models predict future world events such as "will Musk buy twitter," "will a nuclear weapon detonate in Ukraine"? To find out, we're running a competition for retrieval/NLP/reasoning researchers: forecasting.mlsafety.org 🥇place: $100,000 🥈place: $20,000 🥉place: $5,000

Example from the Autocast dataset, including the question, the resolution of the question, and the timeseries of aggregate human expert forecasts (Crowd) from the start date to the time the question resolves. We train a language model to generate forecasts at each timestep, using only news articles available at that timestep (i.e. without allowing any leakage of information from the future).

read image description

ALT

14

59

310

ML Safety

@ml_safety

Oct 26, 2022

“If you cannot measure it, you cannot improve it.” ML Safety research lacks benchmarks. We are offering up to $500,000 in prizes for ML Safety benchmark ideas (or papers). Main site: benchmarking.mlsafety.org Example ideas: benchmarking.mlsafety.org/ideas

2

10

28

Oct 13, 2022

Joe is always worth reading

Quote Tweet

Joe Carlsmith

@jkcarlsmith

Oct 13, 2022

I’m moving my writing from Hands and Cities to a new personal website, joecarlsmith.com. I've also created a substack, joecarlsmith.substack.com, for subscriptions, and a podcast, joecarlsmithaudio.buzzsprout.com, for audio versions of the essays. (1/4)

Show this thread

1

12

Matthew Yglesias

@mattyglesias

As someone who feels pretty uncertain about a lot of the big picture arguments flying around on artificial intelligence, I'm pretty excited to see this from FTX Future Fund. ftxfuturefund.org/announcing-the

28

19

146

Jack Clark

@jackclarkSF

Excellent approach to funding! 'We think X is important and we'll pay you tons of money if you persuade us X is bullshit!' Good chance for people skeptical of these ideas to try and shift a load of capital in a different direction.

Quote Tweet

We're making big bets—a substantial fraction of our capital—on AGI xrisk. But we think it's really possible that we're wrong! Today we're announcing prizes from $15k-$1.5M to change our minds (or those of superforecasters) on AGI. Enter by Dec 23! ftxfuturefund.org/announcing-the

Show this thread

3

13

66

We think it's possible that we're totally out of it on AGI. And we want to learn about it as quickly as because it would change how we allocate 100s of millions of $ (or more). Enter by Dec 23!

3

36

Do you think all of this AGI xrisk stuff is a deluded sideshow? Or do you think that we should be even more focused on AGI? We're launching prizes of up to $1.5M to convince us! (or convince a panel of superforecasters!)

Quote Tweet

We're making big bets—a substantial fraction of our capital—on AGI xrisk. But we think it's really possible that we're wrong! Today we're announcing prizes from $15k-$1.5M to change our minds (or those of superforecasters) on AGI. Enter by Dec 23! ftxfuturefund.org/announcing-the

Show this thread

21

34

231

conversationswithtyler.com

tylercowen

@tylercowen

Aug 10, 2022

My excellent Conversation with Will MacAskill, conversationswithtyler.com/episodes/willi, #philosophy, #effectivealtruism, #EA,

@willmacaskill

William MacAskill on Effective Altruism, Moral Progress, and Cultural Innovation (Ep. 156)

If moral philosophy is a train to crazy town, at what stop should we disembark?

7

17

63

Dustin Moskovitz

@moskov

Aug 10, 2022

Time is running a cover story on Effective Altruism. 🤯

Quote Tweet

TIME

@TIME

Aug 10, 2022

TIME's new cover: Inside the fast-growing movement to do more good in the world ti.me/3zPmLNU

0:15

84.2K views

7

29

323

Aug 6, 2022

I've been finding Squiggle a really useful tool, and I'm excited to see where it's headed. QURI /

@ozziegooen

is hiring — if you're a software developer excited about BOTECs, you could have a big impact there!

Quote Tweet

Ozzie Gooen

@ozziegooen

Aug 3, 2022

We at QURI are releasing an Early Access version of Squiggle. Squiggle is an open-source programming language on JS meant for cost-benefit analysis / risk analysis / probabilistic models. Think of it like the spiritual successor to @getguesstimate. forum.effectivealtruism.org/posts/TfPdb2aM

Show this thread

2

32

Elon Musk

@elonmusk

Aug 2, 2022

Worth reading. This is a close match for my philosophy.

Quote Tweet

William MacAskill

@willmacaskill

Apr 29, 2022

My new book, What We Owe The Future, is now available for pre-order! It makes the case for longtermism, the view that positively affecting the long-run future is a key moral priority of our time. Here's a thread about it...

oneworld-publications.com/what-we-owe-th

basicbooks.com/titles/william

Show this thread

5,015

5,251

45.9K

Without specific countermeasures, the easiest path to transformative AI likely leads to AI takeover...

Jul 19, 2022

Interesting + clarifying post by Ajeya Cotra

alignmentforum.org

I think that in the coming 15-30 years, the world could plausibly develop “transformative AI”: AI powerful enough to bring us into a new, qualitatively different future, via an explosion in science a…

5

1

30

Jul 17, 2022

Really excited that this is out! Great work by Tom Davidson twitter.com/albrgr/status/

This Tweet is unavailable.

13

We're trying to test new funding models; here's some of what we're learning.

1

6

Some example grants:

1

5

Like these regranting anecdotes:

1

2

Some excerpts from our

update in the linked tweet thread! Really excited about the projects we are supporting, the killer team we are building, and some of the things we are working on next (prizes, funder-initiated startups).

Quote Tweet

1/ Yesterday we shared the first update on our work since our launch in late February. Some excerpts in thread! We've made 262 grants and investments so far, totaling ~$132M. ftxfuturefund.org/future-fund-ju

Show this thread

1

20

SBF

@SBF_FTX

Really proud of what the Future Fund team has been working on! ftxfuturefund.org/future-fund-ju

Quote Tweet

An update on our work so far ftxfuturefund.org/future-fund-ju

42

31

277

Also thrilled that

@avitalbalwit

has joined the team! 🎉

1

13

Excited to share what we've done and some of what we're learning along the way!

Quote Tweet

An update on our work so far ftxfuturefund.org/future-fund-ju

1

4

41

Wow!

is a superstar. Very big hire for

@IFP

.

Quote Tweet

Caleb Watney

@calebwatney

Jun 27, 2022

Extremely excited to announce that @heidilwilliams_ is joining @IFP as our Director of Science Policy! Heidi is a superstar in the economics of innovation (and will continue to be — she’s not leaving academia!) We are thrilled that she'll be leading our metascience agenda.

Show this thread

1

6

22

Alexander Berger

@albrgr

Jun 24, 2022

Interesting set of slides from

@ChadJonesEcon

estimating that ~75% of global welfare growth over the past centuries has been population growth rather than income per capita growth: web.stanford.edu/~chadj/slides-

3

9

64

causeexplorationprizes.com

Chris Smith

@Chris_topian

May 25, 2022

“What new cause area should Open Philanthropy consider funding?” Today we’re launching the Cause Exploration Prizes to gather thoughtful answers to this question and others that help us best use our resources

Cause Exploration Prizes

7

68

131

Apollo Academic Surveys

@ApolloSurveys

May 19, 2022

Thanks to

for their support and for inspiring this project.

Quote Tweet

Chris Said

@Chris_Said

May 19, 2022

Announcing a new project: Apollo Academic Surveys. It's like the Chicago IGM poll of economists... but for all academic fields. If you're an academic who is interested in helping organize surveys, DM me. Financial support will be available in many cases. apollosurveys.org

Show this thread

2

18

Matthew Yglesias

@mattyglesias

May 15, 2022

Cash prizes for good new blogs? In my day the humiliation of needing to say the word "blog" out loud was its own reward. effectiveideas.org/#rules

3

8

62

Isabel J.

@brightertmrrw

May 11, 2022

looks like in countries where on average men come even close to doing half of the childcare, disagreement within couples about having another kid is slightly more likely to be due to men not wanting another child rather than women. aeaweb.org/articles?id=10

1

3

14

forum.effectivealtruism.org

Jamie Bernardi

@The_JBernardi

May 4, 2022

A new full time summer programme for undergrads to learn ML safety: AI safety from an empirical, machine learning perspective. Taught by

@DanHendrycks

(UC Berkeley) with lectures from MIT, hugging face & others. Very excited to see this spin up!

Introducing the ML Safety Scholars Program - EA Forum

PROGRAM OVERVIEW The Machine Learning Safety Scholars program is a paid, 9-week summer program designed to help undergraduate students gain skills in machine learning with the aim of using those skil…

5

38

The Blog Prize

@effective_ideas

May 2, 2022

The world in 2072: We want to read about visions of the world 50 years from now. Fiction or forecasts are welcome. Ideally, pieces will be anywhere between 500 and 5000 words. Prize: We’ll award $1000 to the outstanding pieces Deadline: Opens May 1, Closes May 30th

3

8

19

Sam Altman

@sama

Apr 30, 2022

people either happen to the world, or the world happens to them.

62

213

1,690

William MacAskill

@willmacaskill

Apr 29, 2022

My new book, What We Owe The Future, is now available for pre-order! It makes the case for longtermism, the view that positively affecting the long-run future is a key moral priority of our time. Here's a thread about it... 🇬🇧oneworld-publications.com/what-we-owe-th 🇺🇸basicbooks.com/titles/william

390

1,073

5,602

Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.

Anthropic

@AnthropicAI

Apr 29, 2022

We’ve raised $580 million in a Series B. This will help us further develop our research to build usable, reliable AI systems. Find out more:

anthropic.com

Anthropic

14

93

526

Alec Stapp

@AlecStapp

Apr 28, 2022

Tyler Cowen’s talent spotting skill is off the charts 🤯

8

30

473