Executive Order 2023.02 (henceforth "NIGHTMARE WORLD WHERE FACT AND FICTION COEXIST")

  • 🐕 I am attempting to get the site runnning as fast as possible. If you are experiencing slow page load times, please report it.
I think people are overblowing this. There were clips of Jordan Peterson calling for Total Nigger Death years ago. And video effects and motion capture are older than the Internet. All that has happened in the past year or so (and will continue to happen) is make the tech which was once only accessible to big and powerful organizations more readily available for the average person.

Even today, video / audio evidence is not necessarily enough because editing and taking out of context exists. For example, Trump talking about "grabbing them by the pussy" would still not be anywhere near influential without Access Hollywood confirming the tapes are authentic. It basically comes down to who supports the evidence - and in politics, people mostly only believe their side anyways.
Hell, project Veritas came out not too long ago with the recording of that Pfizer employee in which he admitted to some seriously shady stuff, but it has mostly flown under the radar. The people on the left automatically disregard it because it was them who uncovered it and would not bother going into it at all (and the same is true for the Right). It's just selection bias.

TL;DR people will believe what they want to believe already. "It's a deepfake" is just added as another defense for their already existing arsenal.
 
For those of you wondering how to combat an increasingly untrustworthy Internet, look into the Web-of-Trust (WoT) concept. In brief, physically meet people and exchange cryptographic secrets to enable secure communication over untrusted channels. This helps to prevent Sybil attacks and whatnot.

Now, this is no defense against an organization who can successfully trick someone using multiple paid actors and whatnot, but the WoT is perhaps the bare minimum that will be needed for this. Anyway, I'm already part of a WoT and I'm not exactly giving out invitations. Have good luck finding people around yourselves who care about this.
Yeah bro lemme just walk up to kiwi farmers IRL and hand them my crypto keys so we can exchange fart porn once or twice a week, yeah bro that totally is what we need
 
For decades every high-level celebrity, actor, politician, even business executives, have had elaborate voice training to improve their public speaking skills. This typically leads to them have flatter voices, using similar diction and pronunciation, and dialing back any accents or unique quirks so that they'd be more broadly appealing and better understood. Since this has also made them far easier to imitate and now DeepFake, I expect over the next few years we'll see celebs and VIPs shifting the other way, being voice coached into adopting increasingly elaborate cadences or varying inflections/pronunciations as a means of making it harder for AI to copy their voices accurately.

With proper training, humans can radically alter their voices to be unrecognizable or wholly unique. Case in point, David Suchet's famous Poirot voice:


If you've watched any professional impersonators before, you'll quickly find that the hardest part of imitating someone isn't matching their voice, but is getting their quirks of speech and cadence down correctly. Anyone can sound like a New Yorker, but impersonating Trump requires picking the right words to say, talking with hand gestures, cutting off words midway to go on a tangent, gasping for air, etc. Look at all the people who immediately dismiss the Biden video as fake, not because it doesn't match his voice or isn't as blatantly racist as he is, but because it is too well-spoken for a dementia-patient; he hasn't sounded that eloquent in decades.

Something similar could be done to combat DeepFaked videos once AI is good enough to perfectly replicate moving lips and facial animation. Every person already has a slightly unique walk due to small differences in leg length or dexterity; I could see people being trained to constantly change their hand gestures and walking patterns to be even harder for AI to imitate and animate. Who knows, something akin to the Monty Python "Silly Walks" sketch might actually happen for real.

There will also almost certainly arise use of Anti-AI software packages to add in unique changes to audio-video that humans either don't notice or can't (sort of like how TV stations use software to make cuts in their shows to better match up with commercial breaks or add in random black frames to pad out the length of an episode by a second or two). This would actively work to counter AI programming to make it that much more difficult to impersonate accurately, or at least provide a better means of proving that a piece of audio-video has been DeepFaked. I could see security camera companies being the first to implement something like this to prove their footage isn't doctored.
 
I was thinking, there might be a good use for this in on topic threads. A lot of the non-english speaking cows have some really good content that people can't really enjoy. This might be a good way to translate videos.
Good point. In Dankula's series about Drachenlord, he reminds everyone that you have to understand German to truly understand why people hate Rainer so much. Otherwise, he seems like a sympathetic character. English-translated content from him are just the highlights, and him actually speaking English is limited to "HEY! GUY! I... BASTARD YOU!"
 
The potential of AI voice replication and modulation is potentially limitless. Combine this with AI art, AI video generation, AI animation, and AI writing, and we could be in the midst of a brewing AI revolution in terms of content generation, which would completely transform the entertainment industries of the world.

There's a reason why so many are worried about AI content generation, and it's predominantly and primarily commercial and financial in nature. "Ethical" concerns are merely a smokescreen for not wanting to be outperformed, outdone, and eventually (maybe) replaced by machines.

filtered AI is actually the norm in big corpo AI. They'll just make it retarded and also unable to think about bad things. The fact that Eleven hasn't yet is honestly stunning.
Unfiltered open source AI is going to make so much shit obsolete it won't even be funny. We're talking the full decentralization of modern entertainment industries.
 
Last edited:
Just thought I'd share... :lol:
 
@Null Might be worth looking into some of the AI watermarking systems implemented. It wouldn't work for someone running it on their own machines, but a lot of the commercial providers of AI tools add watermarks into images, and maybe audio, that most people couldn't remove or wouldn't be aware of. Just a thought
 
I think people are overblowing this. There were clips of Jordan Peterson calling for Total Nigger Death years ago. And video effects and motion capture are older than the Internet. All that has happened in the past year or so (and will continue to happen) is make the tech which was once only accessible to big and powerful organizations more readily available for the average person.

Even today, video / audio evidence is not necessarily enough because editing and taking out of context exists. For example, Trump talking about "grabbing them by the pussy" would still not be anywhere near influential without Access Hollywood confirming the tapes are authentic. It basically comes down to who supports the evidence - and in politics, people mostly only believe their side anyways.
Hell, project Veritas came out not too long ago with the recording of that Pfizer employee in which he admitted to some seriously shady stuff, but it has mostly flown under the radar. The people on the left automatically disregard it because it was them who uncovered it and would not bother going into it at all (and the same is true for the Right). It's just selection bias.

TL;DR people will believe what they want to believe already. "It's a deepfake" is just added as another defense for their already existing arsenal.
To anyone worrying about the death of truth: photoshop was created in 1987 and using it, people have been aware of the ability to convincingly edit anything, anywhere, for at least ten years now. I'm sure we'll have an equivalent to, "I can tell by the pixels."
Still, I think it does warrant some worry.

All it would take is for someone to use this AI to make it look like you did or said some heinous shit, and at that point, the damage is done: it won't matter whether you can prove it's fake, they'll be people who'll still believe it's real, so you'll have to constantly tell people that it's a deepfake.

I can't imagine how this is NOT going to negatively impact society in the next few years, especially when (and I believe it's only a matter of "when") it improves to the point that it'll take literal experts to decipher whether it's a fake or not. Hell, the AI voices already sound pretty good; I think you could actually trick a few people into believing that a real person said that.

I'm sure it was only barely a year ago when you could tell that an AI made it. Technology is improving too fast.
 
What are some high quality open source ai's I can mess around with? ElvenLabs I know of, though it doesn't let you do custom voices. Voice ai was also recommended to me, but I can't get the thing to work. Any suggestions?
 
  • Dumb
Reactions: K!w!Fartz
What are some high quality open source ai's I can mess around with? ElvenLabs I know of, though it doesn't let you do custom voices. Voice ai was also recommended to me, but I can't get the thing to work. Any suggestions?
Nothing will sadly top ElevenLabs unless you shell out the shekels. You can do the basic premium tier because it won't charge you the first month, just abuse it til you hit the limit and make sure you cancel the sub asap.
 
Its a shame really. While it might be benificial to render the internet completely unfit for messaging with ML, it means that the truly insane will slip through the cracks- especially with the tech improving this fast.
 
So much doomerism here. I got used to talking with people of questionable intelligence and origin long ago when I started shitposting on 4chan. All you are telling me now is that somehow prepared me for what the world was becoming.
 
  • Like
Reactions: Wright
Back