Asher

506 posts

Asher banner
Asher

Asher

@asher5772

@mit, @atlasfellow, emergent ventures // https://t.co/IfTJRdIwoY

เข้าร่วม Kasım 2022
608 กำลังติดตาม953 ผู้ติดตาม
Naomi Bashkansky
Naomi Bashkansky@NaomiBashkansky·
Exciting!!!!
Sam Altman@sama

AI will help discover new science, such as cures for diseases, which is perhaps the most important way to increase quality of life long-term. AI will also present new threats to society that we have to address. No company can sufficiently mitigate these on their own; we will need a society-wide response to things like novel bio threats, a massive and fast change to the economy, extremely capable models causing complex emergent effects across society, and more. These are the areas the OpenAI Foundation will initially focus on, and in my opinion are some of the most important ones for us to get right. The Foundation will spend at least $1 billion over the next year. @woj_zaremba, co-founder of OpenAI, will transition to Head of AI Resilience. I believe that shifting how the world thinks about safety to include a Resilience-style approach is critical, and I am extremely grateful to Wojciech for taking on this role. Wojciech has been my cofounder for the last decade; anyone who knows him will understand what I mean when I say he is one of a kind. He has a lot of ideas about how we build a new kind of AI safety. @JacobTref is joining as Head of Life Sciences and Curing Diseases. @annaadeola, our VP of Global Impact, will transition to Head of AI for Civil Society and Philanthropy. @robert_kaiden is joining as Chief Financial Officer. @jeffarnold is joining as Director of Operations.

English
3
1
57
7.3K
Asher
Asher@asher5772·
@agniv_s calling yourself out
English
1
0
2
232
agniv
agniv@agniv_s·
the hottest men at stanford are stuck in their rooms talking to claude code instead of their crushes
English
12
0
65
6.2K
Asher
Asher@asher5772·
@si_pbc Great work, congrats!!!
English
0
0
6
567
Standard Intelligence
Standard Intelligence@si_pbc·
Computer use models shouldn't learn from screenshots. We built a new foundation model that learns from video like humans do. FDM-1 can construct a gear in Blender, find software bugs, and even drive a real car through San Francisco using arrow keys.
GIF
English
187
400
3.9K
1.1M
Atticus Wang
Atticus Wang@atticuswzf·
Is "a response formatted like this" sometimes better than "a response formatted like this"? To a reward model, yes! RMs are instrumental in shaping model behaviors and alignment. Our paper makes progress uncovering their unexpected preferences. 🧵(1/9)
Atticus Wang tweet media
English
8
12
92
13.8K
Subhash Kantamneni
Subhash Kantamneni@thesubhashk·
We recently released a paper on Activation Oracles (AOs), a technique for training LLMs to explain their own neural activations in natural language. We piloted a variant of AOs during the Claude Opus 4.6 alignment audit. We thought they were surprisingly useful! 🧵
Subhash Kantamneni tweet media
English
11
34
205
26.2K
Asher
Asher@asher5772·
Btw, if it turns out that you can actually shape the pretraining prior using synthetic data, this is a massive fucking deal. Wish we had 10x as many people working on this direction. (I expect that even if it fails, you can produce a bunch of interesting insights).
English
3
0
24
2.2K
Adam Shai
Adam Shai@adamimos·
activation geometry
Adam Shai tweet media
English
13
23
336
17.3K
Paola
Paola@phenoatypical·
i feel a strong urge to increase variance in my life, any suggestions?
English
11
1
24
3.9K
Asher
Asher@asher5772·
@uzpg_ good taste
English
0
0
1
80
Uzay
Uzay@uzpg_·
some favorite non fiction: theory of computation by sipser quantum mechanics by shankar beyond good and evil stranger in the village selfish gene letters to a young poet
English
1
0
7
374
Ryan Greenblatt
Ryan Greenblatt@RyanPGreenblatt·
Inspired by someone wondering how good AIs are at zeroshot CSS animation (like Alex the CSS Husky), I had claude (Opus 4.5, claude code) make a pure CSS animation of a bat without any testing/rendering. It's decent. It's spooky how good AIs are at pure-text drawing/animating.
English
4
1
67
6.8K
claire night skies🪼
claire night skies🪼@clairebookworm1·
for how to make (almost) anything this year i made blonk 🎛️, a music machine/toy that lets you dj with samples & detachable blocks. i believe in tangible art (daws are too Software! too Computer!) & letting us hand-create music without needing to learn the tool™
English
10
2
45
2.8K
jsd
jsd@datagenproc·
Considering creating an Epoch Anki deck
English
5
0
17
221
jsd
jsd@datagenproc·
Do you Anki or another SRS 3 times a week or more?
English
1
0
2
236
Anuja U
Anuja U@heyanuja·
I made a Goodreads for academic papers! (..and blog posts, substacks, lesswrong, etc) Paper Trails [papertrailshq.com] is something I built because I wanted a place where engaging with research felt fun, beautiful, and personal to you I hope you give it a try & love it!
Anuja U tweet media
sanje horah@sanjehorah

i am BEGGING

English
166
1.5K
9.6K
946.8K
Asher
Asher@asher5772·
At NeurIPS this week! Let me know if you're around and want to chat about generalization, interpretability, continual learning, future of AI, etc
English
2
0
12
909
claire night skies🪼
claire night skies🪼@clairebookworm1·
in honor of all the robotics launches this week, this is the robot we made in how to make (almost) anything, the BRAINROTBOT 9000, in just this past week :D
English
3
0
30
2.7K
gavin leech (Non-Reasoning)
Free punishment: Me and the Poseidon lads are going to read every NeurIPS 2025 paper in one sitting. Come help! A party in the NY office, or you can sign up to livestream a tranche from wherever
gavin leech (Non-Reasoning) tweet media
English
7
6
45
4.9K
Asher
Asher@asher5772·
@NinaPanickssery Yep! Leveraging salient concept representations, even when we can't specify a perfect reward signal, seems very useful
English
0
0
0
130