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) →

AI companies aren't really using external evaluators
Zach Stein-Perlman · 2024-05-24T16:01:21.184Z · comments (15)
Laziness death spirals
PatrickDFarley · 2024-09-19T15:58:30.252Z · comments (34)
the case for CoT unfaithfulness is overstated
nostalgebraist · 2024-09-29T22:07:54.053Z · comments (37)
Refusal in LLMs is mediated by a single direction
Andy Arditi (andy-arditi) · 2024-04-27T11:13:06.235Z · comments (93)
[link] I got dysentery so you don’t have to
eukaryote · 2024-10-22T04:55:58.422Z · comments (3)
Believing In
AnnaSalamon · 2024-02-08T07:06:13.072Z · comments (51)
[link] Introducing AI Lab Watch
Zach Stein-Perlman · 2024-04-30T17:00:12.652Z · comments (30)
What are the results of more parental supervision and less outdoor play?
juliawise · 2023-11-25T12:52:29.986Z · comments (31)
AGI Safety and Alignment at Google DeepMind: A Summary of Recent Work
Rohin Shah (rohinmshah) · 2024-08-20T16:22:45.888Z · comments (33)
Modern Transformers are AGI, and Human-Level
abramdemski · 2024-03-26T17:46:19.373Z · comments (88)
MIRI 2024 Mission and Strategy Update
Malo (malo) · 2024-01-05T00:20:54.169Z · comments (44)
SAE feature geometry is outside the superposition hypothesis
jake_mendel · 2024-06-24T16:07:14.604Z · comments (17)
Brute Force Manufactured Consensus is Hiding the Crime of the Century
Roko · 2024-02-03T20:36:59.806Z · comments (156)
LLM Generality is a Timeline Crux
eggsyntax · 2024-06-24T12:52:07.704Z · comments (116)
CFAR Takeaways: Andrew Critch
Raemon · 2024-02-14T01:37:03.931Z · comments (62)
The ‘strong’ feature hypothesis could be wrong
lewis smith (lsgos) · 2024-08-02T14:33:58.898Z · comments (17)
Superbabies: Putting The Pieces Together
sarahconstantin · 2024-07-11T20:40:05.036Z · comments (37)
ChatGPT can learn indirect control
Raymond D · 2024-03-21T21:11:06.649Z · comments (27)
AI Control: Improving Safety Despite Intentional Subversion
Buck · 2023-12-13T15:51:35.982Z · comments (7)
"Slow" takeoff is a terrible term for "maybe even faster takeoff, actually"
Raemon · 2024-09-28T23:38:25.512Z · comments (70)
[link] "How could I have thought that faster?"
mesaoptimizer · 2024-03-11T10:56:17.884Z · comments (32)
Towards more cooperative AI safety strategies
Richard_Ngo (ricraz) · 2024-07-16T04:36:29.191Z · comments (130)
Ability to solve long-horizon tasks correlates with wanting things in the behaviorist sense
So8res · 2023-11-24T17:37:43.020Z · comments (83)
OpenAI: Fallout
Zvi · 2024-05-28T13:20:04.325Z · comments (25)
Mechanistically Eliciting Latent Behaviors in Language Models
Andrew Mack (andrew-mack) · 2024-04-30T18:51:13.493Z · comments (40)
[link] Jaan Tallinn's 2023 Philanthropy Overview
jaan · 2024-05-20T12:11:39.416Z · comments (5)
Pay Risk Evaluators in Cash, Not Equity
Adam Scholl (adam_scholl) · 2024-09-07T02:37:59.659Z · comments (19)
Maybe Anthropic's Long-Term Benefit Trust is powerless
Zach Stein-Perlman · 2024-05-27T13:00:47.991Z · comments (21)
[link] Sam Altman’s Chip Ambitions Undercut OpenAI’s Safety Strategy
garrison · 2024-02-10T19:52:55.191Z · comments (52)
Funny Anecdote of Eliezer From His Sister
Noah Birnbaum (daniel-birnbaum) · 2024-04-22T22:05:31.886Z · comments (6)
This might be the last AI Safety Camp
Remmelt (remmelt-ellen) · 2024-01-24T09:33:29.438Z · comments (34)
Toward A Mathematical Framework for Computation in Superposition
Dmitry Vaintrob (dmitry-vaintrob) · 2024-01-18T21:06:57.040Z · comments (17)
Thoughts on “AI is easy to control” by Pope & Belrose
Steven Byrnes (steve2152) · 2023-12-01T17:30:52.720Z · comments (56)
The impossible problem of due process
mingyuan · 2024-01-16T05:18:33.415Z · comments (64)
[question] Examples of Highly Counterfactual Discoveries?
johnswentworth · 2024-04-23T22:19:19.399Z · answers+comments (100)
Optimistic Assumptions, Longterm Planning, and "Cope"
Raemon · 2024-07-17T22:14:24.090Z · comments (46)
Propaganda or Science: A Look at Open Source AI and Bioterrorism Risk
1a3orn · 2023-11-02T18:20:29.569Z · comments (79)
Response to Aschenbrenner's "Situational Awareness"
Rob Bensinger (RobbBB) · 2024-06-06T22:57:11.737Z · comments (27)
The Sun is big, but superintelligences will not spare Earth a little sunlight
Eliezer Yudkowsky (Eliezer_Yudkowsky) · 2024-09-23T03:39:16.243Z · comments (139)
Self-Other Overlap: A Neglected Approach to AI Alignment
Marc Carauleanu (Marc-Everin Carauleanu) · 2024-07-30T16:22:29.561Z · comments (43)
[link] Sam Altman fired from OpenAI
LawrenceC (LawChan) · 2023-11-17T20:42:30.759Z · comments (75)
How I Learned To Stop Trusting Prediction Markets and Love the Arbitrage
orthonormal · 2024-08-06T02:32:41.364Z · comments (26)
What's Going on With OpenAI's Messaging?
ozziegooen · 2024-05-21T02:22:04.171Z · comments (13)
My AI Model Delta Compared To Christiano
johnswentworth · 2024-06-12T18:19:44.768Z · comments (73)
Two easy things that maybe Just Work to improve AI discourse
jacobjacob · 2024-06-08T15:51:18.078Z · comments (35)
On Not Pulling The Ladder Up Behind You
Screwtape · 2024-04-26T21:58:29.455Z · comments (21)
My Interview With Cade Metz on His Reporting About Slate Star Codex
Zack_M_Davis · 2024-03-26T17:18:05.114Z · comments (187)
OMMC Announces RIP
Adam Scholl (adam_scholl) · 2024-04-01T23:20:00.433Z · comments (5)
A basic systems architecture for AI agents that do autonomous research
Buck · 2024-09-23T13:58:27.185Z · comments (12)
Thinking By The Clock
Screwtape · 2023-11-08T07:40:59.936Z · comments (27)
← previous page (newer posts) · next page (older posts) →