Module 4 / Lesson 12.3
P-Doom Movie
Learning: Watch the 27-minute short film "Writing Doom" (often referred to as the P-Doom movie) embedded below.
Reflection: As you watch the writers debate the season's plot, focus on:
- How the different characters represent the varying perspectives on AI risk (e.g., Optimists, Realists, Interventionists) that we explored previously.
-
The specific analogies used to explain why controlling a superintelligence might be difficult (such as the 5-year-old CEO or Einstein imprisoned by Neanderthals).
In our previous lessons, we discussed the theoretical risks of unaligned Artificial Superintelligence (ASI) and the various policy debates surrounding its development. To help visualize these complex arguments, you will watch a short film that acts as a compelling, narrative-driven exploration of these very topics.
The film follows a group of television writers who are tasked with designing a new TV season where the antagonist is an Artificial Superintelligence. As they try to break the story, they naturally stumble into the core philosophical and technical debates about AI safety, instrumental convergence, and human survival.
Remarks: This film does an excellent job of distilling highly technical AI alignment concepts, like utility functions, instrumental convergence, and deceptive alignment, into accessible arguments. Pay close attention to how quickly a seemingly harmless goal (like winning at chess or predicting tokens) can theoretically lead to catastrophic outcomes if the system is exponentially smarter than its creators.