Sign up to save your podcasts
Or
Share “Untitled Retrospective and Learnings from AI in Context’s First Two VideosDraft” by ChanaMessinger
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
“Untitled Retrospective and Learnings from AI in Context’s First Two VideosDraft” by ChanaMessinger
Note: I used LLMs to draft different parts of this. I've checked almost everything, but there might be some mistakes remaining.
Apologies for posting this on Christmas Eve. I wanted to get this out the door before the end of the year. Questions welcome, and if it's easy to pull metrics to answer them, I will.
Summary80,000 Hours launched a video program in 2025 focused on longform, cinematic, personality-driven content about AI risks. Our first two longform releases were:
- We're Not Ready for Superintelligence (the "AI 2027" video): 8.9M views, ~1.4M watch hours
- If you remember one AI disaster, make it this one (the "MechaHitler" video): 2.7M views, ~419K watch hours
Both videos significantly outperformed our expectations (we'd anticipated 15-50K views for the first). The cost per engagement hour ($0.11 and $0.39 respectively, including staff time) compares favorably to other 80,000 Hours programs.
This post covers: what we spent, what we got, why we think it worked, and what we'd do differently.
The numbersCostsCategoryAI 2027MechaHitlerDirect costs~$50K~$64KStaff hours~450 hrs~450 hrs (Note, I’m assuming it's about the same as for AI 2027, I didn't re-ask people how much time they spent.)Total cost (making some assumptions about we should incorporate staff [...]---
Outline:
(00:34) Summary
(01:33) The numbers
(01:36) Costs
(02:16) Timing
(02:40) Results
(03:46) How valuable is a video watch hour?
(04:24) Qualitative Feedback
(04:28) AI 2027
(05:51) MechaHitler
(06:12) YouTube commenters like:
(06:52) What the comments don't like:
(07:17) Qualitative Analysis
(07:21) Why we think AI 2027 did well
(09:56) Why MechaHitler did less well (but still well)
(10:50) Lessons Learned
(10:54) Overall what we think matters
(11:25) Our guess at what's less important (though we're certainly unsure, maybe if we nailed these, we'd get more success)
(12:24) How our production works
(12:43) The timeline
(13:32) Ideation
(14:06) Scripting
(14:57) Shooting
(15:31) Reshoots / Voiceover
(15:45) Editing
(16:06) Launch
(17:00) What we're still figuring out
(17:36) Closing thoughts
---
First published:
December 24th, 2025
Source:
https://forum.effectivealtruism.org/posts/RCRaBYSqBaMzHzTjF/untitled-retrospective-and-learnings-from-ai-in-context-s
---
Narrated by TYPE III AUDIO.
---
Images from the article:
Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.
View all episodes
By EA Forum Team
4.9
99 ratings
Note: I used LLMs to draft different parts of this. I've checked almost everything, but there might be some mistakes remaining.
Apologies for posting this on Christmas Eve. I wanted to get this out the door before the end of the year. Questions welcome, and if it's easy to pull metrics to answer them, I will.
Summary80,000 Hours launched a video program in 2025 focused on longform, cinematic, personality-driven content about AI risks. Our first two longform releases were:
- We're Not Ready for Superintelligence (the "AI 2027" video): 8.9M views, ~1.4M watch hours
- If you remember one AI disaster, make it this one (the "MechaHitler" video): 2.7M views, ~419K watch hours
Both videos significantly outperformed our expectations (we'd anticipated 15-50K views for the first). The cost per engagement hour ($0.11 and $0.39 respectively, including staff time) compares favorably to other 80,000 Hours programs.
This post covers: what we spent, what we got, why we think it worked, and what we'd do differently.
The numbersCostsCategoryAI 2027MechaHitlerDirect costs~$50K~$64KStaff hours~450 hrs~450 hrs (Note, I’m assuming it's about the same as for AI 2027, I didn't re-ask people how much time they spent.)Total cost (making some assumptions about we should incorporate staff [...]---
Outline:
(00:34) Summary
(01:33) The numbers
(01:36) Costs
(02:16) Timing
(02:40) Results
(03:46) How valuable is a video watch hour?
(04:24) Qualitative Feedback
(04:28) AI 2027
(05:51) MechaHitler
(06:12) YouTube commenters like:
(06:52) What the comments don't like:
(07:17) Qualitative Analysis
(07:21) Why we think AI 2027 did well
(09:56) Why MechaHitler did less well (but still well)
(10:50) Lessons Learned
(10:54) Overall what we think matters
(11:25) Our guess at what's less important (though we're certainly unsure, maybe if we nailed these, we'd get more success)
(12:24) How our production works
(12:43) The timeline
(13:32) Ideation
(14:06) Scripting
(14:57) Shooting
(15:31) Reshoots / Voiceover
(15:45) Editing
(16:06) Launch
(17:00) What we're still figuring out
(17:36) Closing thoughts
---
First published:
December 24th, 2025
Source:
https://forum.effectivealtruism.org/posts/RCRaBYSqBaMzHzTjF/untitled-retrospective-and-learnings-from-ai-in-context-s
---
Narrated by TYPE III AUDIO.
---
Images from the article:
Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.
More shows like EA Forum Podcast (Curated & popular)
View all
Clearer Thinking with Spencer Greenberg
141 Listeners