Who even needs AI generated videos when you can just act out absurdity and pretend it's real?
There's a morbid path from the grainy Iraq war and earlier shaky footage, through IS propaganda which at the time had basically the most intense combat footage ever released to the Ukraine war. Which took it to the morbid end conclusion of endless drone video deaths and edited clips 30+ mins long with day long engagements and defending.
And yes, to answer your belief that there is none - there is loads of "cinematic body cam footage out there now".
I still don't know if it's autogenerated from the original video or recreated manually but yeah it's pretty realistic for the first few seconds.
https://www.theverge.com/2017/12/4/16732912/ufc-video-stream...
Fake images play into that, but they don't need to be AI generated for that to be true, it's been true forever.
I could see it being pretty shocking if I hadn't, but I honestly can't imagine how I'd miss that.
That is also how magicians work, drawing your attention to one particular thing, hiding the secret of the trick from you, sometimes even in plain sight, like in the video.
Or pickpockets, who might bump into you and picking your pocket at the same time, where your attention is focussed on the sudden impact, keeping your attention away from your walled being taken.
> I could see it being pretty shocking if I hadn't, but I honestly can't imagine how I'd miss that.
The point of the video wasn't to count correctly, but to see the gorilla
Essentially, send me a video of something I care about and I will only look for that thing. Most people are not detectives, and even most would-be detectives aren’t yet experts.
At this point, looking at a big tech SoMe feed I would expect that everything is, or at least could be, gen AI content.
The image had all the telltale signs of being AI generated (too much detail, the lights & shadows were the wrong scale, the focus of the lens was odd for the kind of photo, etc). I checked that other group, and sure enough, they claim to be about sharing "miniature dioramas" but all they share is AI-generated crap.
And in the original group, which I'm a member of and is full of people who actually create dioramas -- let's say they are "subject matter experts" -- nobody suspected anything! To them, who are unfamiliar with AI art, the photo was of a real hand-made diorama.
Somebody took AI video editing, and in a year or two, we're already at entire MMA rabbit holes of fake videos.
Commenting mostly as a personal evidence reference of how crazy the World Wide Web has gotten from anecdotal sources.
I won't argue wether text to speech qualifies as an AI but I agree they must be making bank.
But the people who position themselves to profit from the energy consumption of the hardware will profit from all of it: the LLMs, the image generators, the video generators, etc. See discussion yesterday: https://news.ycombinator.com/item?id=41733311
Imagine the number of worthless images being generated as people try to find one they like. Slop content creators iterate on a prompt, or maybe create hundreds of video clips hoping to find one that gets views. This is a compute-intensive process that consumes an enormous amount of energy.
The market for chips will fragment, margins will shrink. It's just matrix multiplication and the user interface is PyTorch or similar. Nvidia will keep some of its business, Google's TPUs will capture some, other players like Tenstorrent (https://tenstorrent.com/hardware/grayskull) and Groq and Cerebras will capture some, etc.
But at the root of it all is the electricity demand. That's where the money will be made. Data centers need baseload power, preferably clean baseload power.
Unless hydro is available, the only clean baseload power source is nuclear fission. As we emerge from the Fukushima bear market where many uranium mining companies went out of business, the bottleneck is the fuel: uranium.
Trial and error content-creation using generative AI, whether or not it creates any real-world value, consumes a lot of electricity.
This electricity demand is likely to translate into demand for nuclear power.
When this demand for nuclear power meets the undersupply of uranium post-Fukushima, higher uranium prices will result.
More supply and lower prices will result.
Not unlike the recent few years in (say) lithium, anticipated demand surged exploration and development, actual demand didn't meet anticipated demand and a number of developed economicly feasible resources were shuttered .. still waiting in the wings for a future pickup in demand.
Look at how long NexGen's Rook 1 Arrow is taking to develop (https://s28.q4cdn.com/891672792/files/doc_downloads/2022/03/...). Spend an hour listening to what Cameco said in its most recent conference call. Look at Kazatomprom's persistent inability to deliver the promised pounds of uranium, their sulfuric acid shortages and construction delays.
Uranium mining is slow and difficult. Existing demand and existing supply are fully visible. There's a gap of 20-40 million pounds per year, with nothing to fill the gap. New mines take a decade or more to develop.
It is not in the slightest like lithium.
Would two decades in global exploration geophysics and being behind the original incarnation of https://www.spglobal.com/market-intelligence/en/industries/m... count?
> Once you know the facts we can talk again.
Gosh - that does come across badly.
When someone compares uranium to lithium, I know I'm not talking to a uranium expert.
All the best to you, and I'll try to be more polite in the future.
Clearly you're a better expert.
> when someone compares uranium to lithium, I know I'm not talking to a uranium expert.
It's about boom bust and shuttering cycles that apply in all resource exploration and production domains.
Perhaps you're a little too literal for analogies? Maybe I'm thinking in longer time cycles than yourself and don't a few years of lag as anything other than a few years.
You are well-prepared to familiarize yourself with the current supply/demand situation. It's time to "make bank", just like you did in 2007... only more so. The 2007 spike was during an oversupplied uranium market and mainly driven by financial actors.
I invite you to begin by listening to any recent interview with Mike Alkin.
Good night and enjoy your weekend.
Might also be an AI voice-changer (i.e. speech2speech) model.
These models are most well-known for being used to create "if [famous singer] performed [famous song not by them]" covers — you sing the song yourself, then run your recording through the model to convert the recording into an equivalent performance in the singer's voice; and then you composite that onto a vocal-less version of the track.
But you can just as well use such a model to have overseas workers read a script, and then convert that recording into an "equivalent performance" in a fluent English speaker's voice.
Such models just slip up when they hit input phonemes they can't quite understand the meaning of.
(If you were setting this up for your own personal use, you could fine-tune the speech2speech model like a translation model, so it understands how your specific accent should map to the target. [I.e., take a bunch of known sample outputs, and create paired inputs by recording your own performances of them.] This wouldn't be tenable for a big low-cost operation, of course, as the recordings would come from temp workers all over the world with high churn.)
That said, this channel has been producing videos well before ChatGPT3.5/4 so at the very least they probably started with human written scripts.
I feel like it might actually be quite complex for AI to pull up the perfect clips and edit them with the script, including timing and everything. Maybe it could be made automatic, but nonetheless it would be a complex process and I don't think possible few years ago. I know Gemini and possibly some others can analyze video if fed to them, but I'm still skeptical that this channel in particular would have done it, when they have always had this frequency of uploads and similar tone.
Also I think there's far better TTS now with ElevenLabs and others so it could be made much more human like.
The only regret I have about that is losing video as a form of evidence. CCTV footage and the like are a valuable tool for solving crimes. That’s going to be out the window soon.
Does this create difficulties if you want to modify the raw video data in any way? Yes it does, even if you just want to save it in a different lossy compression level or format. But these problems aren't insurmountable. Essentially, provenance info can be added for each modification, signed by the entity that made the change, and the end viewer can then decide if they trust the full certificate chain (just as they do now with HTTPS).
Recently someone said here that it's noticable that videos from CCTV cameras are often filmed with a phone or camera on a screen instead of using the original video, and people were speculating that it might be hard or impossible to get access to the original recording because of bureaucracy or something, but that recording a playback on a screen with a phone or camera or something is then often allowed. Maybe they also do this partly so that the original can't be easily messed with by other people.
But yeah if you can verify that a certain video was filmed at a certain time by a certain camera, that is great. Of course the companies providing these cameras should be trustworthy, and that the camera's are actually really sending what they actually record, and that the company itself doesn't mess with the original recordings.
I recall an article posted 1-2 years ago about a camera company (Kodak? Can't remember) which was starting to offer something along these lines.
>the companies providing these cameras should be trustworthy, and that the camera's are actually really sending what they actually record, and that the company itself doesn't mess with the original recordings.
I agree. We can't guarantee any of these things, but on the bright side, the incentives are pointing in the right direction to make self-interested companies choose to behave the right way.
It will complicate things and make the hardware more expensive, so I doubt it will sweep through all consumer camera tech unless the "Is this photo real?" question becomes a crisis. There's also the fact that it would be possible to give individual cameras different private keys, with certificates signed by the manufacturer: This would enable non-repudiation (you would not be able to plausibly deny that you had taken a particular photo/video), which has potentially big upsides but also privacy downsides. I think that could be solved by giving the user the option of signing with their unique camera private key (when the user wants to prove to others that they took the photo themselves) or with the manufacturer's key (when they want to remain anonymous).
Similarly, while Photoshop made it easier it happened at a time where technical advances made the problem harder because everyone’s standards for photos went up dramatically, and so producing a realistic fake was still a slow process for a skilled worker.
Now, it’s increasingly available to everyone and that means that we’re going to see a lot more scams and hoaxes as people without artistic talent or willingness to invest time can make realistic fakes even for minor things. That availability is transformative enough to merit the concern we’ve been seeing here.
The ultimate compression is to reduce the video clip to a latent space vector representation to be rendered on device. :)
Just give us a few more revs of Moore’s law for that to be reasonable.
edit: found a patent… https://patents.google.com/patent/US11388416B2/en