Possible future extrapolation:

In 2-4 years, if we're still alive, anytime you see something IRL this beautiful, your first thought will be to wonder whether it's real or if your VR/AR glasses AI prompt was "beautiful scene of a moth flapping its wings, professional photography, 8k, trending on Twitter"

@TheAllMemeingEye Yeah, I'd believe that.

This Alan Levinowitz tweet opens up an interesting attack surface for this market. Maybe over the next few years we'll actually get better at spotting the uncanny valley!

@JoshuaWilkes I think the answer to Levinovitz's question might be that he thinks it is easy to tell that those images aren't AI because the caption says that they aren't AI

https://media2.giphy.com/media/v1.Y2lkPTc5MGI3NjExMzVmejV1eW1jdWQ3MXNtMXNrZm8zb3p4c2gya21vbGZoMXNvbTZ2ZSZlcD12MV9pbnRlcm5hbF9naWZfYnlfaWQmY3Q9Zw/63jM4dW9fWDtA7BODH/giphy.gif

veo, first attempt

@galaga (it was the first of three, and the best)

@galaga Thank you for confirming that we are currently not in the "casually beat ultra high quality/effort real videos" stage of AI gen, even for series of short clips. Though I wouldn't be surprised by someone who put in a lot of effort being able to do it with AI.

@galaga is there a concise way to specify in the prompt that the wings are being distorted by air resistance due to the high pre-slo-mo speed of wing movement, such that they resemble human sized fabric sheets moving at human speeds? That's really the only thing missing for me

@TheAllMemeingEye dude their wings are clipping into each other. A kindergartener could tell that's AI.

@dgga lol little kids frequently mistake low quality 90s cgi for being real

@galaga looks shit

I have yet to see an AI video that is comparably beautiful to the original video? Not making an argument for a NO resolution, just curious if anyone's seen anything that good from AI.

@jacksonpolack I think the best way to settle this would be for someone to go and actually generate a similar video of months, that would remove much of the ambiguity.

@jacksonpolack I agree. I'm confident we get there, but I think it would be good to post a few specific examples in the comments before a Yes resolution.

honestly I think would be less than 85% (and potentially much less but I haven't thought about it much) if it's based on my standards for beauty, rather than a general standard, but my preferences in things like this are extremely titled towards accurate details and how they combine vs the average. although it might be higher if enough people had preferences similar to mine such that the models were trained with that in mind!

also, I'll take an arbitrary amount of mana at 99

@jacksonpolack Is 85% where you think it is for standards like "can figure it out" or for "first thought" / "cause for serious doubt"? Are those different standards for you?

@ProjectVictory

Edit: I guess I didn't do this right. How can I post a GIF or video?

Back when Sora came out, I tried to make this work. This was by far my best attempt, and I don't think it's close to the quality of the tweet. Maybe newer models would do better.

(I do think this should resolve yes, even if moths or slow-mo or whatever else is hard, many things are at the level)

@journcy Call it? I looked back at the old moth video, and if I was seeing it today, I'd be wondering about AI. (Because of how it's a series of short clips.)

Respectfully, I disagree. Videos created by AI today are not holding up to the original moth video, and certainly people generally aren't all looking at videos posted online and thinking that they're AI. I don't think I have any chance of stopping this market from resolving YES, but I do think that it's a mistake and people who think it should resolve now should examine if they have AI blinders on.

@EliezerYudkowsky I'm currently planning to wait to the end of the year to resolve, just so I'm not doing it based on any one seemingly spectacular clip. But I don't expect my resolution to surprise the market.

@EliezerYudkowsky We're still alive Eliezer

@EliezerYudkowsky I agree that the specific evidence of this video being “a series of short clips” would make savvy viewers wonder about AI video-gen today, but not in a way that seems to line up with the specifics of the original tweet. AI videos today composed from a sequence of short clips without continuity are generally that way because a human has generated multiple videos, and edited together the best bits.

In contrast, to me the tweet seems to imply the AI is able to produce the final output without a human needing to edit it.

@JimHays it does seem to be worded that way. But it also would feel very weird to resolve this market NO if the video quality is there but the model can't to an ffmpeg one-liner

@robm I’m only making a claim about that argument, not about how the market should resolve