Lightnews — Scholar-powered news

Reposted by akbir khan

Epoch AI @epochai.bsky.social · May 8

We’ve added four new benchmarks to the Epoch AI Benchmarking Hub: Aider Polyglot, WeirdML, Balrog, and Factorio Learning Environment!

Before we only featured our own evaluation results, but this new data comes from trusted external leaderboards. And we've got more on the way 🧵

1 2 5

Reposted by akbir khan

Epoch AI @epochai.bsky.social · May 8

4. Factorio Learning Environment by Jack Hopkins, Märt Bakler , and
@akbir.bsky.social

This benchmark uses the factory-building game Factorio to test complex, long-term planning, with settings for lab-play (structured tasks) and open-play (unbounded growth).
jackhopkins.github.io/factorio-lea...

Factorio Learning Environment

Claude Sonnet 3.5 builds factories

jackhopkins.github.io

1 1 3

Reposted by akbir khan

Johannes Gasteiger🔸 @gasteigerjo.bsky.social · Mar 25

New Anthropic blog post: Subtle sabotage in automated researchers.

As AI systems increasingly assist with AI research, how do we ensure they're not subtly sabotaging that research? We show that malicious models can undermine ML research tasks in ways that are hard to detect.

1 3 4

akbir khan @akbir.bsky.social · Mar 18

control is a complimentary approach to alignment.

its really sensible, practical and can be done now, even before systems are superintelligent.

youtu.be/6Unxqr50Kqg?...

Controlling powerful AI

YouTube video by Anthropic

youtu.be

1 4

Reposted by akbir khan

Ethan Mollick @emollick.bsky.social · Feb 25

This is a crazy paper. Fine-tuning a big GPT-4o on a small amount of insecure code or even "bad numbers" (like 666) makes them misaligned in almost everything else. They are more likely to start offering misinformation, spouting anti-human values, and talk about admiring dictators. Why is unclear.

7 43 210

akbir khan @akbir.bsky.social · Feb 11

www.anthropic.com/news/paris-a...

Statement from Dario Amodei on the Paris AI Action Summit

A call for greater focus and urgency

www.anthropic.com

1

akbir khan @akbir.bsky.social · Feb 1

This is the entire goal

Grace @gracekind.net · Jan 31

It’s weird to live in a world where AI models are more aligned than the CEOs of the companies creating them

5

akbir khan @akbir.bsky.social · Jan 30

darioamodei.com/on-deepseek-...

Dario Amodei — On DeepSeek and Export Controls

On DeepSeek and Export Controls

darioamodei.com

1 3

Reposted by akbir khan

Hank Green @hankgreen.bsky.social · Jan 28

The fact that Deepseek R1 was released three days /before/ Stargate means these guys stood in front of Trump and said they needed half a trillion dollars while they knew R1 was open source and trained for $5M.

Beautiful.

Trump announces 500B in AI funding. Five days ago.

400 1.8K 14K

Reposted by akbir khan

Zack Witten @zswitten.bsky.social · Jan 24

Can anyone get a shorter DeepSeek R1 CoT than this?

3 1 17

Reposted by akbir khan

Tom Everitt @tom4everitt.bsky.social · Jan 23

Process based supervision done right, and with pretty CIDs to illustrate :)

1 8

Reposted by akbir khan

Mark Riedl @markriedl.bsky.social · Jan 21

I don’t really have the energy for politics right now. So I will observe without comment:

Executive Order 14110 was revoked (Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence)

2 37 96

akbir khan @akbir.bsky.social · Jan 21

R1 model is impressive

2

Reposted by akbir khan

Sarah Jones @sarahjones.bsky.social · Jan 16

430 6.9K 33K

Reposted by akbir khan

/\__/\__/\__/ @cardiograms.bsky.social · Jan 16

David Lynch smirking in an interview

-Believe it or not, Eraserhead is my most spiritual film.
-Elaborate on that.
-No.

150 4.1K 23K

akbir khan @akbir.bsky.social · Jan 16

fuck the tabloids were right

www.nytimes.com/2025/01/15/t...

She Is in Love With ChatGPT

A 28-year-old woman with a busy social life spends hours on end talking to her A.I. boyfriend for advice and consolation. And yes, they do have sex.

www.nytimes.com

2

Reposted by akbir khan

Ethan Mollick @emollick.bsky.social · Jan 15

New randomized, controlled trial by the World Bank of students using GPT-4 as a tutor in Nigeria. Six weeks of after-school AI tutoring = 2 years of typical learning gains, outperforming 80% of other educational interventions.

And it helped all students, especially girls who were initially behind.

15 89 350

Reposted by akbir khan

Ethan Mollick @emollick.bsky.social · Jan 14

Generative AI has flaws and biases, and there is a tendency for academics to fix on that (85% of equity LLM papers focus on harms)…

…yet in many ways LLMs are uniquely powerful among new technologies for helping people equitably in education and healthcare. We need an urgent focus on how to do that

2 11 69

Reposted by akbir khan

Ethan Mollick @emollick.bsky.social · Jan 14

On one hand, this paper finds adding inference-time compute (like o1 does) improves medical reasoning, which is an important finding suggesting a way to continue to improve AI performance in medicine

On the other hand, scientific illustrations are apparently just anime now arxiv.org/pdf/2501.06458

2 5 71

akbir khan @akbir.bsky.social · Jan 13

my metabolism is noticeably higher in london than the bay.

2

akbir khan @akbir.bsky.social · Jan 10

What can AI researchers do *today* that AI developers will find useful for ensuring the safety of future advanced AI systems? To ring in the new year, the Anthropic Alignment Science team is sharing some thoughts on research directions we think are important.
alignment.anthropic.com/2025/recomme...

Recommendations for Technical AI Safety Research Directions

alignment.anthropic.com

1 7 22

Reposted by akbir khan

Hank Green @hankgreen.bsky.social · Jan 5

My hottest take is that nothing makes any sense at all outside of the context of the constantly increasing value of human life, but that increase in value is so invisible (and exists in a world that was built for previous, lower values) that we constantly think the opposite has happened.

59 89 1.8K

akbir khan @akbir.bsky.social · Jan 5

wait what does that mean?

Does it mean there are bugs in lean, or that it does too much work to check a proof?

akbir khan @akbir.bsky.social · Jan 4

wait isn’t everything just regularisation?

1

Reposted by akbir khan

Zack Witten @zswitten.bsky.social · Jan 4

darioamodei.com/machines-of-...

Dario Amodei — Machines of Loving Grace

How AI Could Transform the World for the Better

darioamodei.com

1 3