Speed arguments against scheming (Section 4.4-4.7 of "Scheming AIs")

This is section 4.4 through 4.7 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

• 0:29 4.4 Speed arguments
• 2:22 4.4.1 How big are the absolute costs of this extra reasoning?
• 7:06 4.4.2 How big are the costs of this extra reasoning relative to the simplicity benefits of
• 9:21 4.4.3 Can we actively shape training to bias towards speed over simplicity?
• 10:27 4.5 The “not-your-passion” argument
• 12:46 4.6 The relevance of “slack” to these arguments
• 13:38 4.7 Takeaways re: arguments that focus on the final properties of the model