LESSWRONG
LW

97
Wikitags

AI Capabilities

Edited by plex last updated 29th Aug 2021

AI Capabilities are the growing abilities of AIs to act effectively in increasingly complex environments. It is often compared to to AI Alignment, which refers to efforts to ensure that these effective actions taken by AIs are also intended by the creators and beneficial to humanity.

Subscribe
Discussion
Subscribe
Discussion
Posts tagged AI Capabilities
137EfficientZero: human ALE sample-efficiency w/MuZero+self-supervised
Ω
gwern
4y
Ω
52
89[Paper] Stress-testing capability elicitation with password-locked models
Ω
Fabien Roger, ryan_greenblatt
1y
Ω
10
61A small update to the Sparse Coding interim research report
Ω
Lee Sharkey, Dan Braun, beren
2y
Ω
5
58Memorizing weak examples can elicit strong behavior out of password-locked models
Ω
Fabien Roger, ryan_greenblatt
1y
Ω
5
299EfficientZero: How It Works
Ω
1a3orn
4y
Ω
50
263Getting 50% (SoTA) on ARC-AGI with GPT-4o
ryan_greenblatt
1y
50
58Competitive programming with AlphaCode
Algon
4y
36
353What DALL-E 2 can and cannot do
Swimmer963 (Miranda Dixon-Luinenburg)
3y
303
278Is AI Progress Impossible To Predict?
alyssavance
3y
39
93Meta AI announces Cicero: Human-Level Diplomacy play (with dialogue)
Jacy Reese Anthis
3y
64
77The case for a negative alignment tax
Cameron Berg, Judd Rosenblatt, Diogo de Lucena, Trent Hodgeson
1y
20
34What will the scaled up GATO look like? (Updated with questions)
Amal
3y
22
16[Crosspost] AlphaTensor, Taste, and the Scalability of AI
jamierumbelow
3y
4
15DeepMind on Stratego, an imperfect information game
sanxiyn
3y
9
10Devil's Advocate: Adverse Selection Against Conscientiousness
lionhearted (Sebastian Marshall)
2y
2
Load More (15/141)
Add Posts