Go offline with the Player FM app!
Does scheming lead to adequate future empowerment? (Section 2.3.1.2 of "Scheming AIs")
Manage episode 387375287 series 3402048
This is section 2.3.1.2 of my report “Scheming AIs: Will AIs fake alignment during training in order to get power?”
Text of the report here: https://arxiv.org/abs/2311.08379
Summary of the report here: https://joecarlsmith.com/2023/11/15/new-report-scheming-ais-will-ais-fake-alignment-during-training-in-order-to-get-power
Audio summary here: https://joecarlsmithaudio.buzzsprout.com/2034731/13969977-introduction-and-summary-of-scheming-ais-will-ais-fake-alignment-during-training-in-order-to-get-power
Chapters
1. Does scheming lead to adequate future empowerment? (Section 2.3.1.2 of "Scheming AIs") (00:00:00)
2. 2.3.1.2 Adequate future empowerment (00:00:33)
3. 2.3.1.2.1 When is the “pay off” supposed to happen? (00:01:22)
4. 2.3.1.2.2 Even if the model’s values survive this generation of training, will they survive long (00:04:20)
5. 2.3.1.2.3 Will escape/take-over be suitably likely to succeed? (00:08:16)
6. 2.3.1.2.4 Will the time horizon of the model’s goals extend to cover escape/take-over? (00:10:05)
7. 2.3.1.2.5 Will the model’s values get enough power after escape/takeover? (00:11:56)
8. 2.3.1.2.6 How much does the model stand to gain from not training-gaming? (00:13:23)
9. 2.3.1.2.7 How “ambitious” is the model? (00:16:38)
10. 2.3.1.3 Overall assessment of the classic goal-guarding story (00:21:43)
56 episodes
Manage episode 387375287 series 3402048
This is section 2.3.1.2 of my report “Scheming AIs: Will AIs fake alignment during training in order to get power?”
Text of the report here: https://arxiv.org/abs/2311.08379
Summary of the report here: https://joecarlsmith.com/2023/11/15/new-report-scheming-ais-will-ais-fake-alignment-during-training-in-order-to-get-power
Audio summary here: https://joecarlsmithaudio.buzzsprout.com/2034731/13969977-introduction-and-summary-of-scheming-ais-will-ais-fake-alignment-during-training-in-order-to-get-power
Chapters
1. Does scheming lead to adequate future empowerment? (Section 2.3.1.2 of "Scheming AIs") (00:00:00)
2. 2.3.1.2 Adequate future empowerment (00:00:33)
3. 2.3.1.2.1 When is the “pay off” supposed to happen? (00:01:22)
4. 2.3.1.2.2 Even if the model’s values survive this generation of training, will they survive long (00:04:20)
5. 2.3.1.2.3 Will escape/take-over be suitably likely to succeed? (00:08:16)
6. 2.3.1.2.4 Will the time horizon of the model’s goals extend to cover escape/take-over? (00:10:05)
7. 2.3.1.2.5 Will the model’s values get enough power after escape/takeover? (00:11:56)
8. 2.3.1.2.6 How much does the model stand to gain from not training-gaming? (00:13:23)
9. 2.3.1.2.7 How “ambitious” is the model? (00:16:38)
10. 2.3.1.3 Overall assessment of the classic goal-guarding story (00:21:43)
56 episodes
All episodes
×Welcome to Player FM!
Player FM is scanning the web for high-quality podcasts for you to enjoy right now. It's the best podcast app and works on Android, iPhone, and the web. Signup to sync subscriptions across devices.