Live data from Hacker News

South Park creators have new political satire series with AI-generated deepfakes

theregister.com

61–70 of 313 posts

Re: South Park creators have new political satire series with AI-generated deepfakes

#61

I understand the worries behind deep fakes, but i'm going to be a bit facetious here, hasn't getting the most realistic, lifelike images always been the goal of computer graphics? Wouldn't deepfakes be the ultimate culmination of lifelike CGI. So lifelike it's indistinguishable from the real thing? Like for most of my life, this is the goal I keep hearing about. Now it's here, suddenly it's too real. I dunno just see…

I don't think anyone is disputing the idea that technology in this arena can/will/should advance, but rather that this specific set of technologies come with consequences that are, to many of us, absolutely terrifying.

I also don't think we truly understood how susceptible so many people are to made up nonsense until recent years, so this particular kind of made up nonsense suddenly looks like a serious threat to social stability and cohesion.

Re: South Park creators have new political satire series with AI-generated deepfakes

#62

This feels like the start of something very bad - they are going to be upfront about the deepfakes... others are not. It’s going to get to the point where you can’t even say in a court room “well we have video of him doing this”. The fact that deepfakes will exist will erode confidence in even those things that are true. At the same time it will add additional fake situations to the conversation. Worst part is that e…

Actually, I'm a bit more optimistic about this. One thing that worries me is that everything is becoming immutable, saved forever, and digitally signed. Hard to claim you didn't post something if your account is super secure. You can't go anywhere without cameras taking your picture. And if someone steals your bitcoin or your smart contract has a mistake, your money is gone. You can't argue with an algorithm. And the…

Everything is not immutable or even becoming immutable. Content is merely endpoints. Unless extraordinary care is taken, like publishing hashes, the response from the endpoints can change over time. Lots of content has changed, for instance on Wikipedia. In theory you could wade through the wiki history. I don't know if that is protected by hashing or not. Lots of content gets shoved down the memory hole.

Re: South Park creators have new political satire series with AI-generated deepfakes

#63
post #21

Earlier quoted context omitted.

True, but photoshop is usually done by hand. An AI-generated deepfake could maybe eventually be trained so that it outputs videos that are indistinguishable from the real thing.

The detection side also gets the benefits of AI though. I can't see the generation side getting that far ahead.

From what I understand though, generators always have an advantage because the generator is allowed to "see" the discriminator's gradients during training. [0]

>The model for the discriminator is usually more complex than the generator (more filters and more layers) and a good discriminator gives quality information. In many GAN applications, we may run into bottlenecks where increasing generator capacity shows no quality improvement. Until we identify the bottlenecks and resolve them, increasing generator capacity does not seem to be a priority for many partitioners. [1]

Put another way: GAN training ends when the discriminator can no longer meaningfully distinguish real from fake. By definition then, the best generator will have no useful discriminator that can distinguish its output from real data. (conversely, if you did have such a discriminator, you could use it to train a better generator)

[0] https://developers.google.com/machine-learning/gan/generator

[1] https://towardsdatascience.com/gan-ways-to-improve-gan-perfo...

Re: South Park creators have new political satire series with AI-generated deepfakes

#64
In terms of using deep fakes to impersonate political figures, I do think we'll ultimately be okay on that front. I can see the entertainment industry being flipped on it's head and fabricated conspiracy theories hitting a fever pitch but with initiatives like CAI* cropping up, my hope is that there will be a digital wax seal of sorts to make sure what you're seeing is what the author intended.

* https://contentauthenticity.org

Re: South Park creators have new political satire series with AI-generated deepfakes

#66
post #24

Talking TV heads are already deep fakes. I'm glad computer fakes will finally erode the confidence of public into those people, from news anchors to presidents. Took almost a century to arrive here.

While I agree on your first point, the rest completely misses that the public will continue to be tribal and instead only believe the talking heads of their given tribe as authentic in a world rife with deep fakes.

I fully expect the outcome to be increased tribalism, when faith in your tribe's leadership and information sources is the only source of comfort, confidence, and "truth".

I don't think it will be the opposite where everything is questioned and critical thinking suddenly becomes prolific as you seem to imply.

Re: South Park creators have new political satire series with AI-generated deepfakes

#67

This feels like the start of something very bad - they are going to be upfront about the deepfakes... others are not. It’s going to get to the point where you can’t even say in a court room “well we have video of him doing this”. The fact that deepfakes will exist will erode confidence in even those things that are true. At the same time it will add additional fake situations to the conversation. Worst part is that e…

Yes, deepfakes are going to kill plausible deniability in audio and video.

> deepfakes are going to kill plausible deniability in audio and video

Huh? Don't deepfakes do the exact opposite - make plausible deniability in audio and video a much stronger argument?

Before if you denied a video of you was genuine that would not be plausible. Now it would be plausible.

Re: South Park creators have new political satire series with AI-generated deepfakes

#69
Let's remember their politics, though - they used to market themselves as "both sides are bad" indpendants/libertarians, but now they're full-out Trumpers.

I could see clips from this show making the rounds on lower-information parts of social media like Facebook that are already prone to being hoodwinked by conservative fake news like Alex Jones and the like. Professionally-produced deepfakes becoming the exact attack on democracy that other people in this thread are proposing will "innoculate" us.

Re: South Park creators have new political satire series with AI-generated deepfakes

#70
post #7

I _think_ that the Michael Cain section is Peter Serafinowicz doing a Michael Cain impression (which has then been deepfaked), but it could be Lyre Bird or similar! The fact that I really can't make my mind up which it could be and the weird sense of doubt, to me, means that this has hit the mark perfectly. I've been thinking about this video a lot since I first saw it. It's genuinely very unsettling.

Wouldn't be surprised. While searching him on Youtube I even stumbled on Peter Serafinowicz's appearance on Stephen Colbert where they're talking about him even doing 'bad lip reading' videos for trump in different personalities, one of them being 'Sassy Trump', so he might have had something to do with this show coming about in general.

I don't think he's doing the voice for Sassy Justice, though. That sounds more like Trey Parker to me. But they probably brought him in to do Michael Caine.

https://youtu.be/Oe28UOFNlxk?t=138

Post reply on HN