AI ALIGNMENT FORUMTags
AF

AI Capabilities

EditHistorySubscribe
Discussion (0)
Help improve this page (1 flag)
EditHistorySubscribe
Discussion (0)
Help improve this page (1 flag)
AI Capabilities
Random Tag
Contributors
4plex

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.

Posts tagged AI Capabilities
Most Relevant
6
49EfficientZero: human ALE sample-efficiency w/MuZero+self-supervised
gwern
6mo
19
2
58EfficientZero: How It Works
1a3orn
5mo
1
1
31Interpreting Yudkowsky on Deep vs Shallow Knowledge
Adam Shimi
5mo
2
1
42The alignment problem in different capability regimes
Buck Shlegeris
8mo
12
1
29PaLM in "Extrapolating GPT-N performance"
Lukas Finnveden
1mo
13
1
30OpenAI Solves (Some) Formal Math Olympiad Problems
Michaël Trazzi
4mo
0
1
5Epistemic Strategies of Safety-Capabilities Tradeoffs
Adam Shimi
7mo
0
0
75DeepMind: Generally capable agents emerge from open-ended play
Daniel Kokotajlo
10mo
11
0
25Testing PaLM prompts on GPT3
Yitzi Litt
1mo
0
1
29AI Tracker: monitoring current and near-future risks from superscale models
Edouard Harris, Jeremie Harris
6mo
7
0
15Estimating training compute of Deep Learning models
Lennart Heim, Jaime Sevilla, Marius Hobbhahn, Tamay Besiroglu, Anson Ho
4mo
4
1
7What’s the backward-forward FLOP ratio for Neural Networks?
Marius Hobbhahn, Jaime Sevilla
5mo
0
1
8How to measure FLOP/s for Neural Networks empirically?
Marius Hobbhahn
6mo
0
Add Posts