Resolves as YES if an AI generated video has reached 10 million views on YouTube before January 1st 2025
Different number of views:
/RemNi/will-an-ai-generated-video-reach-10-9511a17489a9 (this question)
/RemNi/will-an-ai-generated-video-reach-20-261fe41fe4dc
/RemNi/will-an-ai-generated-video-reach-50-5580e28e7d88
/RemNi/will-an-ai-generated-video-reach-10
/RemNi/will-an-ai-generated-video-reach-20
/RemNi/will-an-ai-generated-video-reach-50
/RemNi/will-an-ai-generated-video-reach-1b
/RemNi/will-an-ai-generated-video-reach-2
/RemNi/will-an-ai-generated-video-reach-5
Different years:
/RemNi/will-an-ai-generated-video-reach-10-9511a17489a9 (this question)
/RemNi/will-an-ai-generated-youtube-video-9d632ac5c2f1
/RemNi/will-an-ai-generated-youtube-video-dffa37ad4ea8
Different platforms:
/RemNi/will-an-ai-generated-video-reach-10-9511a17489a9 (this question)
/RemNi/will-an-ai-generated-tiktok-video-r
/RemNi/will-an-ai-generated-instagram-vide
/RemNi/will-an-ai-generated-x-video-reach
/RemNi/will-an-ai-generated-facebook-video
Strict version:
/RemNi/will-an-ai-generated-youtube-video
Other questions:
/RemNi/will-we-get-agi-before-2025
At least 90% of the frames and audio in the video must be generated by an AI, but an editing process involving a human is permissible within the context of this question. The video must be at least 45 seconds long (YouTube shorts are acceptable provided they fulfil this criteria).
Intros and outros are not considered as "part of the video" unless these are also AI-generated. If a video has 5 seconds of intros and outros, then the video must be at least 50 seconds long in order to include 45 seconds of AI-generated content.
In order for the video to fulfil the criteria of "AI generated", a neural network (or some similar class of model) must be significantly involved in the production of each frame counted in the 45 seconds of AI-generated content in the video.
The "main subject" of the video must be AI generated in order to qualify. For example, a music video with a real singer superimposed on an AI generated background would not qualify. However, an AI generated singer in a virtual scene on which real backup dancers have been superimposed would qualify.
For a frame to be considered AI generated, then the majority of the pixels must have been produced by a process that involved a neural network (or similar architecture) to a substantial degree.
Silent videos are acceptable, as are videos with a still image, provided this image is AI generated. If the video is not silent then the audio must be broadly generated by an AI. If a video has an audio track that contains both AI generated (synthetic) and recorded audio (e.g a real human's voice), then this media qualifies only if the AI generated audio content is the main "subject" of the audio track. For example, a real human voice singing over synthetic AI-generated background music would not qualify, except in the case that the voice was only heard in a few seconds of the soundtrack.
Background music or sound effects that are not AI-generated are permissible provided that 90% of the video contains AI-generated audio on top (e.g. AI-generated voices), and that the AI-generated audio is the "main subject" of the audio track (cannot be simply AI-generated sound effects on top of a music track created by a human).
Related questions
@RemNi I want to check whether AI Michiko's 3 Reasons to Go Braless (5M views) qualify except for view count.
https://www.youtube.com/watch?v=cWnL_N3XfdM
@SanghyeonSeo i'd say it's similar to the Balenciaga video in that the music is not ai generated (I think), and is doing the... heavy lifting... In many parts of the audio track.
That being said the ai generated voice does cover much more of the audio track than the Balenciaga video. Think it reaches 90% of total the track in this case?
@RemNi if it's gets closer to 10M views then someone will have to measure the fraction of the audio track with the voiceover. I suspect it's just under 90% in this case because of the pauses when going from part 1 to part 2 etc when just the background music is playing.
However, if that background music is ai generated then it definitely qualifies
@traders I'm inclined to resolve this question as YES since the Harry Potter Balenciaga video has gone above 10M views: https://youtu.be/iE39q-IKOzA?si=Fg6ZEPfjxiogyXB9
EDIT: I found the soundtrack for the background music: https://www.youtube.com/watch?v=memDWTAvcXs
After some thought, since this music is a pre-existing soundtrack, and the voices are present in less than 80% of the video, I'm of the opinion now that this video does in fact not meet the criteria of "At least 90% of the audio in the video must be generated by an AI", and therefore does not qualify. But it does come very close.
Happy to hear any further input from others
@RemNi I won't object to the resolution, but I confess: this thing never registered in my head as an "AI generated video" because I kept thinking of it as a glorified slide-show put together by a human. (When I read the word "video", I don't expect that a static image displayed for 10 seconds will qualify, though if it's on youtube, I can't really argue that it doesn't count as a "youtube video".)
@Tasty_Y Yes, I'm considering creating a stricter version of these markets which can exist alongside. I think that's the best solution. Then traders who are more interested in the stricter criteria can vote on that one.
It is still only January 2024 so there's lots of time for a "*strictly* AI generated video" to become popular on YouTube before 2025.
@Tasty_Y Here is the strict version: https://manifold.markets/RemNi/will-an-ai-generated-youtube-video