LessWrong 2.0 Reader

View: New · Old · Top

Restrict date range: Today · This week · This month · Last three months · This year · All time

← previous page (newer posts) · next page (older posts) →

New paper shows truthfulness & instruction-following don't generalize by default
joshc (joshua-clymer) · 2023-11-19T19:27:30.735Z · comments (0)
[link] Are There Examples of Overhang for Other Technologies?
Jeffrey Heninger (jeffrey-heninger) · 2023-12-13T21:48:08.954Z · comments (50)
Approaching Human-Level Forecasting with Language Models
Fred Zhang (fred-zhang) · 2024-02-29T22:36:34.012Z · comments (6)
[link] shoes with springs
bhauth · 2023-12-30T21:46:55.319Z · comments (6)
AI #48: Exponentials in Geometry
Zvi · 2024-01-18T14:20:07.869Z · comments (9)
Understanding SAE Features with the Logit Lens
Joseph Bloom (Jbloom) · 2024-03-11T00:16:57.429Z · comments (0)
LessOnline Festival Updates Thread
Ben Pace (Benito) · 2024-04-18T21:55:08.003Z · comments (26)
[link] More people getting into AI safety should do a PhD
AdamGleave · 2024-03-14T22:14:48.855Z · comments (24)
Does AI risk “other” the AIs?
Joe Carlsmith (joekc) · 2024-01-09T17:51:47.020Z · comments (3)
[link] Towards shutdownable agents via stochastic choice
EJT (ElliottThornley) · 2024-07-08T10:14:24.452Z · comments (7)
The Sense Of Physical Necessity: A Naturalism Demo (Introduction)
LoganStrohl (BrienneYudkowsky) · 2024-02-24T02:56:31.458Z · comments (1)
[link] Pacing Outside the Box: RNNs Learn to Plan in Sokoban
Adrià Garriga-alonso (rhaps0dy) · 2024-07-25T22:00:55.398Z · comments (8)
What's next for the field of Agent Foundations?
Nora_Ammann · 2023-11-30T17:55:13.982Z · comments (23)
On the Latest TikTok Bill
Zvi · 2024-03-13T18:50:05.398Z · comments (7)
[question] Shane Legg's necessary properties for every AGI Safety plan
jacquesthibs (jacques-thibodeau) · 2024-05-01T17:15:41.233Z · answers+comments (12)
Managing catastrophic misuse without robust AIs
ryan_greenblatt · 2024-01-16T17:27:31.112Z · comments (17)
Paper out now on creatine and cognitive performance
Fabienne · 2023-11-26T10:58:29.745Z · comments (2)
Woods’ new preprint on object permanence
Steven Byrnes (steve2152) · 2024-03-07T21:29:57.738Z · comments (1)
A hermeneutic net for agency
TsviBT · 2024-01-01T08:06:30.289Z · comments (4)
[link] Against Nonlinear (Thing Of Things)
tailcalled · 2024-01-18T21:40:00.369Z · comments (18)
[link] "Why I Write" by George Orwell (1946)
Arjun Panickssery (arjun-panickssery) · 2024-04-25T16:02:28.668Z · comments (2)
The LessWrong 2022 Review: Review Phase
RobertM (T3t) · 2023-12-22T03:23:49.635Z · comments (7)
Consider the humble rock (or: why the dumb thing kills you)
pleiotroth · 2024-07-04T13:54:15.593Z · comments (11)
[link] microwave drilling is impractical
bhauth · 2024-06-12T22:16:00.199Z · comments (14)
[link] Talk: "AI Would Be A Lot Less Alarming If We Understood Agents"
johnswentworth · 2023-12-17T23:46:32.814Z · comments (3)
[link] Sam Altman, Greg Brockman and others from OpenAI join Microsoft
Ozyrus · 2023-11-20T08:23:00.791Z · comments (15)
SAEs (usually) Transfer Between Base and Chat Models
Connor Kissane (ckkissane) · 2024-07-18T10:29:46.138Z · comments (0)
Rationalists are missing a core piece for agent-like structure (energy vs information overload)
tailcalled · 2024-08-17T09:57:19.370Z · comments (9)
Memorizing weak examples can elicit strong behavior out of password-locked models
Fabien Roger (Fabien) · 2024-06-06T23:54:25.167Z · comments (5)
Aligned AI is dual use technology
lc · 2024-01-27T06:50:10.435Z · comments (31)
[link] An Opinionated Evals Reading List
Marius Hobbhahn (marius-hobbhahn) · 2024-10-15T14:38:58.778Z · comments (0)
Apply to ESPR & PAIR, Rationality and AI Camps for Ages 16-21
Anna Gajdova (anna-gajdova) · 2024-05-03T12:36:37.610Z · comments (5)
How you can help pass important AI legislation with 10 minutes of effort
ThomasW · 2024-09-14T22:10:50.386Z · comments (2)
We Inspected Every Head In GPT-2 Small using SAEs So You Don’t Have To
robertzk (Technoguyrob) · 2024-03-06T05:03:09.639Z · comments (0)
[link] Announcing the $200k EA Community Choice
Austin Chen (austin-chen) · 2024-08-14T00:39:37.350Z · comments (8)
The Problem With the Word ‘Alignment’
peligrietzer · 2024-05-21T03:48:26.983Z · comments (8)
Mira Murati leaves OpenAI/ OpenAI to remove non-profit control
Sodium · 2024-09-25T21:15:17.315Z · comments (4)
Against empathy-by-default
Steven Byrnes (steve2152) · 2024-10-16T16:38:49.926Z · comments (22)
Voting Results for the 2022 Review
Ben Pace (Benito) · 2024-02-02T20:34:59.768Z · comments (3)
Measurement tampering detection as a special case of weak-to-strong generalization
ryan_greenblatt · 2023-12-23T00:05:55.357Z · comments (10)
On the UBI Paper
Zvi · 2024-09-03T14:50:08.647Z · comments (6)
[link] This is Water by David Foster Wallace
Nathan Young · 2024-04-24T21:21:09.445Z · comments (16)
AI Alignment Research Engineer Accelerator (ARENA): Call for applicants v4.0
James Fox · 2024-07-06T11:34:57.227Z · comments (7)
Referendum Mechanics in a Marketplace of Ideas
Martin Sustrik (sustrik) · 2024-08-25T08:30:01.901Z · comments (2)
Transfer Learning in Humans
niplav · 2024-04-21T20:49:42.595Z · comments (1)
[link] Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning
Dan Braun (Daniel Braun) · 2024-05-17T16:25:02.267Z · comments (10)
[question] What's the theory of impact for activation vectors?
Chris_Leong · 2024-02-11T07:34:48.536Z · answers+comments (12)
[link] [EAForum xpost] A breakdown of OpenAI's revenue
dschwarz · 2024-07-10T18:09:20.017Z · comments (5)
AI #86: Just Think of the Potential
Zvi · 2024-10-17T15:10:06.552Z · comments (8)
Some negative steganography results
Fabien Roger (Fabien) · 2023-12-09T20:22:52.323Z · comments (5)
← previous page (newer posts) · next page (older posts) →