The worst thing about deepfakes is that we know about them

Originally published at: https://boingboing.net/2019/10/16/the-worst-thing-about-deepfake.html

1 Like

I know, right?!

2 Likes

right now the more tangible threat is how the idea of deepfakes can be invoked to make the real appear fake

That’s always seemed the far likelier outcome. When you combine that with an utterly shameless public figure who is already able to convince his base that up is down and anyone who tells you different is fake news, it’s a recipe for disaster. Just looking at what the lack of a shared reality of objective fact is already doing to our system of government, and I’m pessimistic about the future.

6 Likes

We’ve already got Republicans denying objective reality (even in situations when they previously accepted it), so it’s not like things really change with the ability to plausibly deny what you see. It’ll just reinforce that dynamic, cementing it place going forward as the “new normal” rather than some temporary madness by conservatives.

Even more likely outcome: deep fakes being used in situations where you don’t expect them, in scams. Which is already happening. The erosion of reality on a personal level will probably be more significant, because unlike rejecting consensus reality for political reasons, it will be a new phenomenon that impact us on a daily basis.

6 Likes

Also, the best thing about DeepFakes is we know about them.

How do we trust Photographs in the digital era, when Photoshop is so accessible?

Simple: We don’t. We evaluate the photograph. Depending on the importance of the photograph, we decide how hard we need to scrutinize it. We know that photographs can be easily faked.

Experts can look at the photograph and see if it looks faked to them. They can tell by the pixels, generally. :wink:

More seriously, cameras now have the option of signing a hash for the photo to show that it came from the camera undoctored.

We just need to adopt these technologies and attitudes with respect to video as well; and the good thing is that we have a window to adopt them before people get hurt by fakes. We won’t have people found guilty and sentenced by fake video. We will know to evaluate videos and confirm that they are real. We understand that they can be faked, and faked cheaply: which is a good thing, because anything we can do, the government can do better.

6 Likes

It sounds like Roko’s Basilisk is just a warmed over version of Pascal’s Wager. Which Ive never found very compelling. High stakes cannot cancel out low risk.

But it does say quite a lot about how people conceptualize authority in the abstract. When I resolve Pascal’s wager, I am assigning more weight to my own conscience than to that of some hypothetical deity or very real church. Not everyone is going to subscribe to that kind of epistemology, and that’s something to consider when stepping outdoors.

Something similar is happening with digital media. The ability and willingness to fake visual documents means we need to rethink our ideas about authority. Just because a digital proxy is more convenient, doesnt make it better, just look at paper ballots.

It makes sense to me that democracy should be the first thing we notice under fire from weaponized disinformation, but its a problem older than the printing press. I remember the cover story in Whole Earth Review that claimed their photorealistic image of flying saucers over city streets meant that every other reprinted photo could now be doubted.

Consensus reality needs time to arrive at a correct understanding. Electronic media tend to take that time away, and favor quick lies over slow truths. Its going to take some conscientious work to reclaim that time. The high stakes may not involve being tortured by a robot, but a livable atmosphere.

7 Likes

I very much agree with you. :slight_smile:

I think part of it is that even in the analogue era, there was a lot of photo manipulation going on, but a lot of people never questioned photographs because everyone knew “You can’t fake a photo” even though you could.

With video, we have major motion pictures and we don’t think twice about them showing impossible stuff because special effects. But we don’t consider that the same techniques could be used on body camera or security camera footage. So we don’t authenticate it.

And as you say, 100% - we need time to do this, and we don’t have it, and we need to reunderstand authority and authenticity.

ETA: Also paper ballots are the only way to secure democracy, you are quite correct in that too! :slight_smile:

1 Like

Skimmed the boring stuff, but when is Barack Panther going to be released?

7 Likes

Yeah I disagree. It is better that we know there is danger and can guard against it, even if that makes us paranoid. It’s like not being told that a lion has escaped the zoo. Would you approve of the choice not to tell people because there is a pretty small chance of you running in to it?

1 Like

“There is nothing either good or bad, but thinking makes it so.”

The future is fun.

I don’t think ignorance is the story. Instead, they are stating that the very fact that we know deep fakes exist means that it can undermine truth.

A recording doesn’t even have to be a fake for someone to claim it is fake, people will believe whoever is on their side.

It’s fake recordings all the way down, even if all of them are real.

It’s the kind of eye-rolling sophistry that gets widely circulated only because people are entertained by fake proofs of silly conclusions, and because the argument itself is too annoying for anyone to care to spend the time to refute it. Like the Ontological Argument, or the “proof” that the sum of positive integers is -1/12, except that those are at least moderately interesting conversation starters.

It’s beside the point anyhow, because there is no detectable link between Roko’s Basilisk and this argument about deepfakes, unless they are literally the only two examples of abstract reasoning you’ve ever seen.

The point about deepfakes is reasonable – the belief that you should never trust your eyes is indeed more dangerous than the possibility of actually being tricked in a small proportion of instances. Where I disagree is with the implicit assumption that we’re all, like, helpless tube worms wiggling around on the ocean floor, hoping that the morsels our superiors drop into our gaping mouths will be The Truth.

It’s become apparent that this is exactly what a certain type of commentator believes (of their audience, at least). I’ve been seeing a lot of “how do we know what to believe any more?” think-pieces from people who obviously mean “how will the proles know what to believe if they stop venerating the opinions I shart out between lunches?” Says more about you than your audience, buddy.

From my personal perspective, stuff like deepfakes is not all that scary because, oh, what, you mean I can no longer know what to think by uncritically absorbing anything I see on any given screen? OH NO WHATEVER SHALL I DO LOL.

Also… um… it does remain to be seen whether software will produce a “deepfake” that is convincing to people born after 1960. We may be over-privileging the concerns of the Gulliblest Generation here.

4 Likes

If I understand you, then, there’s two levels of deepfake. There’s the stuff so compelling, it can fool even one such as yourself. And then there’s stuff that’s just barely good enough to fool fox news and its viewers. The first category doesn’t yet exist (lets assume) and the second category… I can’t tell whether or not you think its a problem worth discussing.

In a less fucked up world, I wouldn’t be concerned about those who can’t afford to consider Roko’s Basilisk without falling down the rabbit hole. But when one such got into the white house, I got to worrying about it.

In any case, USian propaganda rarely tells the audience what to think. Pretty much always, it tells the audience what their neighbors think. And when we are thouroughly convinced that the viewers to our left and right are significantly stupider and less worthy than ourselves, then the conviction springs up out of our subconscious as if we thought of it ourselves- there is no point forming a co-op, joining a union, starting an affinity group, banding together… when its obvious that everyone is so dumb (excepting ourselves) that they deserve this chaos. We can’t fix this problem because its self inflicted.

Im not sure how best to push back against this strategy, but Im reluctant to give my enemies such an easy win.

No, I was just pointing out, in passing, that we’re slightly ahead of ourselves in assuming totally convincing synthetic video will ever exist.

But leaving that aside, I just meant that video lying is not essentially different to the regular lying that people have dealt with for as long as language has existed. Put it this way: I can quote MLK saying “I have a goose”, but that flawless text deepfake doesn’t shatter the very concept of reality.

If I fake a video of Bernie Sanders doing a sieg heil, how is that any different? Why do we assume network news would report that as reality, any more than they’d pass on an anonymous email describing the same thing in words?

IMO the answer is, we don’t actually assume anything of the sort; we’re just pretending to talk about a new kind of lying, in order to suggest that certain audiences are literally incapable of critical thought, to a degree that would be obviously absurd if we came out and said it like that.

In my experience, faked video evidence need not be all that convincing, in order to be effective.

During the hunt for Osama Bin Laden, US authorities released a video they said was their guy claiming credit for the 9/11 attacks. It was… not just a poor quality video, but obviously manipulated. The murky shadows all seemed to be in exactly the most convenient places, there was a squirrelly quality to it that I was reminded of when I saw the felt skin faces of Space_Cowboys.

And for as much noise at the 9/11 truthers were making about cell calls at altitude and WTC7, I heard not a peep about this film’s authenticity. Whatever scam was being perpetrated at the time, was pretty successful, Id say.

A similarly fishy audio recording was released in the aftermath of the Branch Davidian fire. Supposedly they had a bug inside the compound, that caught the voices of the people inside, frantically setting up incendiary booby traps just as the federal forces were starting to move in… yet apparently this intel didn’t keep them from being completely surprised by the outcome!

(What was really telling about this one, was people talking the way no one does in real life. Only in radio plays have I heard that kind of narration of the action)

…and then, if you like, theres the great grandaddy of faked evidence, the bogus “magic bullet” recovered from a hospital gurney after JFK’s murder. It fooled no one, but in the end, it didn’t matter.

Ultimately, its not the quality of derpfaked video that’s the danger -I think I’m in agreement with you there.- Rather, its the temptation this capability offers to those who should know better, to use it to justify actions they will totally get away with regardless.

Yeah I get that. If you know that lion is on the loose, every squirrel rooting in the bushes, every dark shadow, each time you open the door, the lion is lurking. Still better to know the lion is out there then not to know. If you don’t know that deep fakes are out there, you cannot look out for them. The basilisk can (only) be avoided by NOT knowing about it.

I think the relevant question is whether is will ever be real, because at present it seems not to be. but the conclusion still follows /pedantry

@anansi133 I think the key difference is that Pascal’s Wager is about what may or may not exist, and the most obvious problem with it is privileging an a priori exorbitantly unlikely specific hypothesis. In contrast, Roko’s Basilisk is a thing that could be invented, within the universe we actually inhabit, by the effort of humans. The thought experiment contains a causal loop that Pascal’s wager lacks.

Edit to add: it’s also part of a small but interesting body of thought about whether things that don’t actually exist can and should compel parts of our decision making processes. I enjoy this short story on acausal trade as another example.

“Gosh, there’s a robot designed to torture me unless I build it. I guess I better take that design, and make it real so I don’t get tortured”

Said no one, ever. At least no one I’d trust with a soldering iron.

1 Like

“God could make a Hell and torture me in it for eternity”

is not really different from

“Magic Robot could make a Simulation and torture me in it for eternity”

I keep looking for the glue that makes these kinds of conspiracies so appealing…

Conservatives part ways with liberals in this country, mostly around how dangerous the world truly is. Conservatives say you have to be tough to survive such a hazardous world, even if (especially!) the world becomes more dangerous from your choices.

Where finding objective truth comes into play, science types, and those sympathetic to science, kind of grok that the real truth is difficult to root out, and deliberately obscuring the truth, sets everybody back a little bit.

Those who reflexively call “bullshit!” on video evidence they dont want to accept, are most likely projecting their own dishonesty. If a deepfaked video is the only evidence for something having happened, there will be other evidence out there that will will contradicr it. Journalists have made this process almost as reliable as the sciency one.

Its not the mere existence of deepfakes that causes the problem, its the laziest possible interpretation of “democracy” that insists a vote is a vote, no matter how corrupt or ill informed it is.