Joe Carlsmith Audio
Total duration:
31 h 49 min
AI for AI safety
Joe Carlsmith Audio
27:51
Paths and waystations in AI safety
Joe Carlsmith Audio
18:07
When should we worry about AI power-seeking?
Joe Carlsmith Audio
46:54
What is it to solve the alignment problem?
Joe Carlsmith Audio
40:13
How do we solve the alignment problem?
Joe Carlsmith Audio
08:43
Fake thinking and real thinking
Joe Carlsmith Audio
78:47
Takes on "Alignment Faking in Large Language Models"
Joe Carlsmith Audio
87:54
(Part 2, AI takeover) Extended audio from my conversation with Dwarkesh Patel
Joe Carlsmith Audio
127:33
(Part 1, Otherness) Extended audio from my conversation with Dwarkesh Patel
Joe Carlsmith Audio
238:38
Introduction and summary for "Otherness and control in the age of AGI"
Joe Carlsmith Audio
12:23
Second half of full audio for "Otherness and control in the age of AGI"
Joe Carlsmith Audio
251:02
First half of full audio for "Otherness and control in the age of AGI"
Joe Carlsmith Audio
187:29
Loving a world you don't trust
Joe Carlsmith Audio
63:54
On attunement
Joe Carlsmith Audio
44:14
On green
Joe Carlsmith Audio
75:13
On the abolition of man
Joe Carlsmith Audio
69:22
Being nicer than Clippy
Joe Carlsmith Audio
47:30
An even deeper atheism
Joe Carlsmith Audio
25:12
Does AI risk "other" the AIs?
Joe Carlsmith Audio
13:15
When "yang" goes wrong
Joe Carlsmith Audio
21:32
Deep atheism and AI risk
Joe Carlsmith Audio
46:59
Gentleness and the artificial Other
Joe Carlsmith Audio
22:39
In search of benevolence (or: what should you get Clippy for Christmas?)
Joe Carlsmith Audio
52:52
Empirical work that might shed light on scheming (Section 6 of "Scheming AIs")
Joe Carlsmith Audio
28:00
Summing up "Scheming AIs" (Section 5)
Joe Carlsmith Audio
15:46
Speed arguments against scheming (Section 4.4-4.7 of "Scheming AIs")
Joe Carlsmith Audio
15:19
Simplicity arguments for scheming (Section 4.3 of "Scheming AIs")
Joe Carlsmith Audio
19:37
The counting argument for scheming (Sections 4.1 and 4.2 of "Scheming AIs")
Joe Carlsmith Audio
10:40
Arguments for/against scheming that focus on the path SGD takes (Section 3 of "Scheming AIs")
Joe Carlsmith Audio
29:03
Non-classic stories about scheming (Section 2.3.2 of "Scheming AIs")
Joe Carlsmith Audio
24:34
Does scheming lead to adequate future empowerment? (Section 2.3.1.2 of "Scheming AIs")
Joe Carlsmith Audio
22:54
The goal-guarding hypothesis (Section 2.3.1.1 of "Scheming AIs")
Joe Carlsmith Audio
19:11
How useful for alignment-relevant work are AIs with short-term goals? (Section 2.2.4.3 of "Scheming AIs")
Joe Carlsmith Audio
09:21
Is scheming more likely if you train models to have long-term goals? (Sections 2.2.4.1-2.2.4.2 of "Scheming AIs")
Joe Carlsmith Audio
09:01
"Clean" vs. "messy" goal-directedness (Section 2.2.3 of "Scheming AIs")
Joe Carlsmith Audio
16:44
Two sources of beyond-episode goals (Section 2.2.2 of "Scheming AIs")
Joe Carlsmith Audio
21:25
Two concepts of an "episode" (Section 2.2.1 of "Scheming AIs")
Joe Carlsmith Audio
12:08
On "slack" in training (Section 1.5 of "Scheming AIs")
Joe Carlsmith Audio
07:12
Situational awareness (Section 2.1 of "Scheming AIs")
Joe Carlsmith Audio
09:27
Why focus on schemers in particular? (Sections 1.3-1.4 of "Scheming AIs")
Joe Carlsmith Audio
31:17
