The OTHER AI Alignment Problem: Mesa-Optimizers and Inner Alignment
Skills:
AI Alignment Basics90%Reading ML Papers80%Research Methods70%AI Safety Engineering60%AI Ethics & Policy50%
This "Alignment" thing turns out to be even harder than we thought.
# Links
The Paper: https://arxiv.org/pdf/1906.01820.pdf
Discord Waiting List Sign-Up: https://forms.gle/YhYgjakwQ1Lzd4tJ8
AI Safety Career Bottlenecks Survey: https://www.guidedtrack.com/programs/n8cydtu/run
Mesa-optimizers: https://aisafety.info/questions/8160/
What is deceptive alignment: https://aisafety.info/questions/8EL6/
Inner vs. outer alignment: https://aisafety.info/questions/8428/
Adversarial training: https://aisafety.info/questions/935A/
# Referenced Videos
Intelligence and Stupidity - The Orthogonality Thesis: http://youtu.be/hEUO6pjwFOo
9 Examples of Specification Gaming: https://youtu.be/nKJlF-olKmg
Why Would AI Want to do Bad Things? Instrumental Convergence: https://youtu.be/ZeecOKBus3Q
Hill Climbing Algorithm & Artificial Intelligence - Computerphile: http://youtu.be/oSdPmxRCWws
AI Gridworlds - Computerphile: http://youtu.be/eElfR_BnL5k
Generative Adversarial Networks (GANs) - Computerphile: http://youtu.be/Sw9r8CL98N0
# Other Media
The Simpsons Season 5 Episode 19: "Sweet Seymour Skinner's Baadasssss Song"
1970s Psychology study of imprinting in ducks. Behaviorism: http://youtu.be/2xd7o3z957c
With thanks to my excellent Patreon supporters:
https://www.patreon.com/robertskmiles
- Timothy Lillicrap
- Gladamas
- James
- Scott Worley
- Chad Jones
- Shevis Johnson
- JJ Hepboin
- Pedro A Ortega
- Said Polat
- Chris Canal
- Jake Ehrlich
- Kellen lask
- Francisco Tolmasky
- Michael Andregg
- David Reid
- Peter Rolf
- Teague Lasser
- Andrew Blackledge
- Frank Marsman
- Brad Brookshire
- Cam MacFarlane
- Jason Hise
- Phil Moyer
- Erik de Bruijn
- Alec Johnson
- Clemens Arbesser
- Ludwig Schubert
- Allen Faure
- Eric James
- Matheson Bayley
- Qeith Wreid
- jugettje dutchking
- Owen Campbell-Moore
- Atzin Espino-Murnane
- Johnny Vaughan
- Jacob Van Buren
- Jonatan R
- Ingvi Gautsson
- Michael Greve
- Tom O'Connor
- Laura Olds
- Jon Halliday
- Paul Hobbs
- Jeroen De Dauw
- Lupuleasa Ion
Watch on YouTube ↗
(saves to browser)
Sign in to unlock AI tutor explanation · ⚡30
Playlist
Uploads from Robert Miles AI Safety · Robert Miles AI Safety · 31 of 47
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
▶
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
Predicting AI: RIP Prof. Hubert Dreyfus
Robert Miles AI Safety
Respectability
Robert Miles AI Safety
Are AI Risks like Nuclear Risks?
Robert Miles AI Safety
Avoiding Negative Side Effects: Concrete Problems in AI Safety part 1
Robert Miles AI Safety
Avoiding Positive Side Effects: Concrete Problems in AI Safety part 1.5
Robert Miles AI Safety
Empowerment: Concrete Problems in AI Safety part 2
Robert Miles AI Safety
Why Not Just: Raise AI Like Kids?
Robert Miles AI Safety
Reward Hacking: Concrete Problems in AI Safety Part 3
Robert Miles AI Safety
The other "Killer Robot Arms Race" Elon Musk should worry about
Robert Miles AI Safety
Reward Hacking Reloaded: Concrete Problems in AI Safety Part 3.5
Robert Miles AI Safety
What Can We Do About Reward Hacking?: Concrete Problems in AI Safety Part 4
Robert Miles AI Safety
What can AGI do? I/O and Speed
Robert Miles AI Safety
AI learns to Create ̵K̵Z̵F̵ ̵V̵i̵d̵e̵o̵s̵ Cat Pictures: Papers in Two Minutes #1
Robert Miles AI Safety
AI Safety at EAGlobal2017 Conference
Robert Miles AI Safety
Scalable Supervision: Concrete Problems in AI Safety Part 5
Robert Miles AI Safety
Superintelligence Mod for Civilization V
Robert Miles AI Safety
Why Would AI Want to do Bad Things? Instrumental Convergence
Robert Miles AI Safety
Experts' Predictions about the Future of AI
Robert Miles AI Safety
AI Safety Gridworlds
Robert Miles AI Safety
Friend or Foe? AI Safety Gridworlds extra bit
Robert Miles AI Safety
Safe Exploration: Concrete Problems in AI Safety Part 6
Robert Miles AI Safety
Why Not Just: Think of AGI Like a Corporation?
Robert Miles AI Safety
How to Keep Improving When You're Better Than Any Teacher - Iterated Distillation and Amplification
Robert Miles AI Safety
Is AI Safety a Pascal's Mugging?
Robert Miles AI Safety
AI That Doesn't Try Too Hard - Maximizers and Satisficers
Robert Miles AI Safety
Training AI Without Writing A Reward Function, with Reward Modelling
Robert Miles AI Safety
9 Examples of Specification Gaming
Robert Miles AI Safety
10 Reasons to Ignore AI Safety
Robert Miles AI Safety
Sharing the Benefits of AI: The Windfall Clause
Robert Miles AI Safety
Quantilizers: AI That Doesn't Try Too Hard
Robert Miles AI Safety
The OTHER AI Alignment Problem: Mesa-Optimizers and Inner Alignment
Robert Miles AI Safety
Deceptive Misaligned Mesa-Optimisers? It's More Likely Than You Think...
Robert Miles AI Safety
Intro to AI Safety, Remastered
Robert Miles AI Safety
We Were Right! Real Inner Misalignment
Robert Miles AI Safety
Apply to AI Safety Camp! #shorts
Robert Miles AI Safety
Win $50k for Solving a Single AI Problem? #Shorts
Robert Miles AI Safety
Free ML Bootcamp for Alignment #shorts
Robert Miles AI Safety
Apply Now for a Paid Residency on Interpretability #short
Robert Miles AI Safety
Why Does AI Lie, and What Can We Do About It?
Robert Miles AI Safety
Apply to Study AI Safety Now! #shorts
Robert Miles AI Safety
AI Ruined My Year
Robert Miles AI Safety
Learn AI Safety at MATS #shorts
Robert Miles AI Safety
Using Dangerous AI, But Safely?
Robert Miles AI Safety
AI Safety Career Advice! (And So Can You!)
Robert Miles AI Safety
Robot Dog! Unitree Go2 review #shorts #robot #dog
Robert Miles AI Safety
Tech is Good, AI Will Be Different
Robert Miles AI Safety
Apply for the Affine Superintelligence Alignment Seminar #shorts
Robert Miles AI Safety
More on: AI Alignment Basics
View skill →
🎓
Tutor Explanation
DeepCamp AI