We’re underestimating the mind-warping potential of fake video • Hacker News Board

Re: We’re underestimating the mind-warping potential of fake video

#71

Post by Cyphase » Mon, Jul 23, 2018, 8:22 PM UTC

Felz wrote: post #50

ZeroBugBounce wrote: post #24

Earlier quoted context omitted.

So is not agreeing on facts/truth The Great Filter?

Hm, I'd say not agreeing on facts/truth is more of a byproduct of not agreeing on what we want. As a civilization, we still have the fatal problem that people want what's best for themselves, and we don't have great ways of making them work towards a common good. Worse still, as issues become more complex there are more ways to twist policy towards self-interest while maintaining a facade of group-interest. Evolution…

> I don't think this is The Great Filter though. More likely our civilization disintegrates until it reaches a level of complexity we can handle again, but it's not going to be totally wiped out.

I think people assume it often, but I don't think the Great Filter actually requires destruction of the species/civilization. Maybe it _is_ nukes, or bioweapons; or maybe, the Great Filter (or _a_ Great Filter) is #7 on Hanson's list[0], "Tool-using animals with big brains". So there could be plenty of species out there that use tools and even have civilization, but not enough or the right kind of brainpower for developing space travel. So the Great Filter could be a rigid wall that prevents further progress towards space travel, or an elastic trampoline that keeps pushing civilizations back whenever they get too close to it (what you said). Or it could be a wood chipper. We can only speculate.

[0] https://en.wikipedia.org/wiki/Great_Filter#The_Great_Filter

[note] I say "space travel", but this also includes remotely-detectable signs of intelligence (e.g. signals, or Dyson spheres), and if our remote sensing gets good enough, of life.

Re: We’re underestimating the mind-warping potential of fake video

#72

Post by laurex » Mon, Jul 23, 2018, 8:28 PM UTC

What this article is getting at is that memory and "truth" are not really compatible concepts, which I think most neuroscience supports. Research around vision generally shows that even what we see, first hand (not in video) is an assemblage or construction, not a "record." Yet humans in general tend to believe that if they see something, it's either real or "faked" (like magic) rather than suspect because of the natural limitations of our brain system (which are also advantages for survival). We also have an even harder time to distinguish 'reality' when our emotions are involved. "Fake" video does present more of an issue, I think, than past "fake media", because we simply have more emotional response to it than other media.

Re: We’re underestimating the mind-warping potential of fake video

#73

Post by PeterisP » Mon, Jul 23, 2018, 8:33 PM UTC

olefoo wrote: post #4

I suspect it will be quite confusing for a while; but once people have been cheated a few times because of faked video it will become something else. We will all start picking up on peoples kinematics and timing. There is an opportunity for some type of chain of custody app that guarantees the integrity of footage from camera to screen...

I'm not so sure if "we will all start picking up on peoples kinematics and timing" is going to happen - if anything, we should expect the video faking systems to adapt faster to this than masses of people; it currently has some issues with kinematics just because the concept is quite new and unpolished, give it an extra year or two, and those systems will pick up (and reproduce) the president's kinematics far better than I do.

Re: We’re underestimating the mind-warping potential of fake video

#74

Post by ChainOfFools » Mon, Jul 23, 2018, 8:34 PM UTC

SubiculumCode wrote: post #63

billylindeman wrote: post #31

Earlier quoted context omitted.

I think audio/video authentication is gonna be huge also. One thought would be merkle proof of hashes of the keyframes and then a signature of the merkle root. drop that bad boy into an eos / eth contract.

I think it may be the true use case for crypto.

Sybil problem still remains. blockchain history can be rewritten relatively cheaply if defense is controlled by an army of nodes under obfuscated control of a single entity or clique. even worse when blind faith is put into authenticity of such information.

Re: We’re underestimating the mind-warping potential of fake video

#75

Post by PeterisP » Mon, Jul 23, 2018, 8:37 PM UTC

politician wrote: post #60

octosphere wrote: post #19

We’re also underestimating the mind-warping potential of detecting fake video via carefully trained algorithms.

Why not both? Generative Adversarial Networks http://blog.aylien.com/introduction-generative-adversarial-n...

The particular videos are designed by this approach (GANs), and thus the generator is highly adapted to generate data that can't be distinguished from a real video with these techniques.

If it's built with state-of-art GANs, then you don't be able to reliably detect it with discriminators from these state-of-art GANs. If you manage to build a better automatic discriminator than the authors did, then they could directly use it in the adversarial approach to build a better generator, and your detection approach would again cease functioning.

Remember that in the end there's a fundamental asymmetry - in theory, there could be a perfect generator (not that we're close to perfect yet) that generates undistinguishable samples, and thus there can not be a perfect discriminator.

Re: We’re underestimating the mind-warping potential of fake video

#76

Post by andrewmcwatters » Mon, Jul 23, 2018, 8:47 PM UTC

SubiculumCode wrote: post #64

politician wrote: post #59

My personal belief is that we should be generating vast amounts of deep fake videos and audio clips now that the technology is available. We should intentionally and hastily move to pollute the information streams around us as rapidly as possible in order to force an evolution in how we process information. To attempt to slow down this arms race, or to ignore its implications just puts us at a further disadvantage to…

That's step 1. Step 2 is to provide a product that resolves the issue through crypto/blockchain certification.

What does this mean? Can you elaborate?

Re: We’re underestimating the mind-warping potential of fake video

#77

Post by gowld » Mon, Jul 23, 2018, 8:49 PM UTC

makomk wrote: post #39

ethbro wrote: post #25

Earlier quoted context omitted.

The Vox article is also making a nuanced point -- that memory is constructed subconsciously. So the insidious thing? Creating 100 videos with only slight political position adjustments, and thereby (even if the videos are later revealed as fakes!) shifting a voter's "gut" impression of a candidate. I'd be amazed if we don't start seeing audio & video cryptographic authentication in the next 10 years. The subject of a…

This is probably already happening with good old fashioned Photoshop (or not even that sometimes). For instance, when a Telegraph journalist mistook a photoshopped photo of Trump with a white supremacist for the real deal and tweeted it with the comment that "this isn't complicated"[1], I doubt many people who saw the original tweet even saw the correction[2]. Over 28,000 retweets for the original claim, under a thou…

One feature missing is a way to mark every account that tweeted the original but not the correction as an unreliable possible-troll.

Re: We’re underestimating the mind-warping potential of fake video

#78

Post by yifanl » Mon, Jul 23, 2018, 8:55 PM UTC

politician wrote: post #59

My personal belief is that we should be generating vast amounts of deep fake videos and audio clips now that the technology is available. We should intentionally and hastily move to pollute the information streams around us as rapidly as possible in order to force an evolution in how we process information. To attempt to slow down this arms race, or to ignore its implications just puts us at a further disadvantage to…

So a form of targeted accelerationism? That's an interesting thought, but assuming it works as intended, wouldn't the end result of this be a total destruction of trust?

A whole society where all non-physical interaction is treated with total suspicion can't be a pleasant one to live in. Maybe that's the end result either way though.

Re: We’re underestimating the mind-warping potential of fake video

#79

Post by gowld » Mon, Jul 23, 2018, 8:58 PM UTC

Spivak wrote: post #35

Miredly wrote: post #22

Earlier quoted context omitted.

I'm sorry, but your example falls a little flat when you look at how Fox News became the lynchpin of the shit-show the US is sinking in to right now. People adapted to it, sure- but that's the problem.

If a single TV station had the power you describe you would think they might be a little more successful at using it by now. Not saying they have no effect, but if the station never existed I don't think our history would be drastically different.

What's make you think they are unsuccessful?

Sure, a TV station can't annoint a President of whatever. The power to destroy is a weaker power than the power to create, but is still an immense power.

Rupert Murdoch's lifelong business philosophy is that all information gatekeepers should be destroyed, so that no one's opinion (or facts) are elevated over any other's. It's pure libertarian-democracy of ideas that throws the baby (education and critical thinking) out with the bathwater (suppression of outsider voices). The modern "Internet of Trolls" is the culmination of his philosophy.

https://www.theguardian.com/media/greenslade/2011/jul/18/reb...

Re: We’re underestimating the mind-warping potential of fake video

#80

Post by gowld » Mon, Jul 23, 2018, 9:00 PM UTC

denimalpaca wrote: post #26

A good way to fight this is to cryptographically sign videos/audio. Might be a bit computationally expensive compared to a text message, but for very high-profile people, having a copy of the video/audio and it's cryptographic signature will be absolute proof against fakes. We don't even need AI algorithms to learn how to detect fakes in this case, it just becomes a matter of comparing hashes.

How does the average user know whose signatures to trust? If Facebook censors videos by signatures, people will scream censorship and ignore the signatures.