This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
Willy Wonka Has Done It Again
Well, we all knew this was coming. We had AI that could describe the progress of a scene. We had AI that could generate photorealistic images. Of course creating a sequence of images that form a logically coherent scene would be possible. Perhaps most surprising thing is how not surprising it is, and yet, it still feels like magic.
Looking back to my television after watching so many Sora videos is uncanny. Somehow, my brain has already updated subconsciously to, "these figures onscreen might be fake." Will I eventually have the same impression for those I meet in real life? Those complaining about "misinformation" in relation to the 2024 election aren't necessarily wrong, but they are failing at the task of extending these logical conclusions beyond their own personal preoccupations. How can one be concerned about Joe Biden deepfakes in a world where the Experience Machine is being built in plain sight?
While the technological achievement of Sora is highly impressive, the faux Japanese texts within the generated Tokyo scenes were clearly fake. I can read Japanese hiragana, katakana and also a fair amount of Chinese characters, and most of the pseudo-Japanese text was nothing but garbled nonsense mixed in with a few legitimate characters, some of which were mirrored or missing strokes. This fits a general pattern I've noticed in many AI-generated images that happen to contain passages of text, which is almost always made-up gibberish with no basis in reality, as though the AI is treating it as a mere pattern to imitate.
Interestingly, the same thing happens to text in (lucid) dreams.
Back when DallE2 was the cutting edge, somebody suggested that the images being produced might suggest that to the Extent DallE has any 'internal experience' it probably feels roughly like the equivalent to a lucid dream to it.
Probably not true but it sticks in my head nonetheless.
More options
Context Copy link
More options
Context Copy link
To be fair, I believe this is standard practice in the film industry for at least some applications, and AI may just be following conventions from its training data.
More options
Context Copy link
More options
Context Copy link
Deep fakes have existed for a while now, but they tend to be easily detected . The moving or transposed parts tend to not align well with the static parts and audio, creating a weird effect. For enough money the quality can be improved. I imagine a well funded actor could create a deep fake which passes such scrutiny. There was a notable incident: $25 million stolen due to a deep fake. This will get worse, and likely will require better ways of vetting sources as deep fake technology improves.
A reversion to holding meetings face-to-face, and relying on handwritten documents and physical stamps/seals of authenticity seems in the cards right now.
More options
Context Copy link
More options
Context Copy link
It's not just concerns about election interference:
I wonder who the "read teamers" are. I wonder what third party people or organizations/NGOs are being consulted. I know it's too much to hope for any transparency from OpenAI on this front.
With all of Sam Altman's professional notoriety, it's notable that his Twitter bio is nothing more than the Star of David, and he identifies strongly as Jewish although it is not clear if he literally believes in Yahweh.
Will Sora continue the OpenAI trend of being tuned to be highly defensive of Jewish identity and Zionism, and critical of white identity and nationalism? Last month Altman said he believes antisemitism “is a significant and growing problem in the world", so we should all expect Sora to be especially tuned to fight against or otherwise prevent antisemitism.
Now that the prospect of Plato's Cave being projected by endlessly-generated AI content is getting closer, these questions are more important than ever.
More options
Context Copy link
I'm a little suspicious, the videos seem too good, so I wonder if there was any human 'tidying-up' involved? Or how many prompts did Sora go through to get the finished article? Maybe it really is as amazing as it seems, but I do wonder.
Almost certainly cherrypicked, but sam a. has been sharing videos of people's prompts on X.
https://x.com/sama/status/1758220311735181384?s=20
https://x.com/sama/status/1758218820542763012?s=20
https://x.com/sama/status/1758218059716939853?s=20
https://x.com/sama/status/1758206987094147252?s=20
https://x.com/sama/status/1758200420344955288?s=20
Some of 'em are bad, some are decent. Still an improvement over what came before.
Some of those are definitely better than others, the ocean bicycle race one doesn't get the prompt quite right (why is there a dolphin flying in the sky?) but it's not bad if these are indeed immediate responses to prompts.
Not as good and flawless as the claims, but not bad. Still visibly computer-generated (the hamster's legs are all wrong) but give them time and they may indeed produce "indistinguishable from created by humans".
More options
Context Copy link
More options
Context Copy link
It looks like they are using the Unreal Engine or something similar to generate synthetic training data.
More options
Context Copy link
Obviously there was plenty of room for trickery, but Altman was apparently fielding random requests on Twitter and sharing them out with a few minutes of turnaround after the launch announcement. There was likely plenty of selection going on and of course it's possible that the requests were preplanned and posted by plants, but OpenAI's track record suggests that this will be available for the public to play with before long so any major shenanigans seem unlikely to me.
More options
Context Copy link
More options
Context Copy link
It’s a pretty impressive advance; there were a few short (30-60) frame models earlier, but they were extremely prone to weird or trippy physics or perspective faults. There’s def some selection going on, but from the paper, SORA seems able to produce more coherent videos than even many traditional 3d animation.
The most interesting part for me is going to be the potential for transfer or vid2vid. There’s a number of videos in the paper along those lines, and while they don’t always work (pixel art), I could def see a lot of animators making rough skeletons of the fundamentals of a scene with 3d tools (maybe with physics simulations for larger objects/fluid) or live actors, sending it through SORA, and then repeating modification to get fine details down.
More options
Context Copy link
More options
Context Copy link