From the BBC:
The number one sign you're watching an AI videoQuoteIn the last six months, AI video-generators got so good that our relationship with cameras is about to melt. Here's the best-case scenario: you'll get fooled, over and over again, until you're so fed up that you question every single thing you see.
https://www.bbc.com/future/article/20251031-the-number-one-sign-you-might-be-watching-ai-video
From Duck Soup (1933):
Teasdale: But I saw you with my own eyes.
Chicolini: Well, who ya gonna believe me or your own eyes?
https://quoteinvestigator.com/2018/07/31/believe-eyes/
I suppose eye witnesses will be more important when video evidence is so easy to generate.
>eye witnesses
Houston, we have a problem....
Why eyewitnesses fail - PMC
https://pmc.ncbi.nlm.nih.gov/articles/PMC5544328/
Eyewitness Errors - How mental processes change our memories. | Psychology Today
https://www.psychologytoday.com/us/blog/the-forensic-view/202310/eyewitness-errors
Right, but we maybe won't really have an alternative? I suspect it won't be long until the average person will lose faith in audio and video content being genuine.
Isnt there a move to verify video? Have it from a source you trust who can verify it.
>average person will lose faith in audio and video content being genuine.
Like that "1 to 3 min gap" outside Epstein's cell? And remember the "grassy knoll" with JFK never really went away, it's just the people most interested in it got too old or died off and that was way before digital or AI.
>verify
We're doomed when audio or video becomes no longer accepted as evidence in a court of law and that challenge is coming sooner than we think.
I wonder if there won't be some method equivalent to a checksum hash that verifies a file has not been altered.
Of course, that won't be sufficient, because an AI file isn't necessarily altered. But there could be something like a given camera would have a salt and then the file would have a hash based on the device salt, the timestamp and the content of the video. Then you could verify that a video from a given camera had not been altered.
Of course, at that point, you could no long edit video, so it would only be useful in situations like a court of law, not, say, a documentary film. But documentary film makers could potentially have a third-party audit.
It would, of course, require new algorithms, a new platform and all-new hardware, but we did it once already for payments over the web and, as we have proven, that has gone perfectly and they are uncrackable.
Quote from: ergophobe on November 07, 2025, 02:55:06 AMthat has gone perfectly and they are uncrackable.
Funny :)
A part of it is the volume of film, it cannot all be verified. So it comes back to trust. Do you Trust the BBC for example to provide un doctored video.
Generally, if it is on their website, I do. If the same message is coming from other verified sources, then yes, that adds to it.
If it is supplied by a political party, then no.
So lets hope the Political "elite" don't get their hands on our most reliable sources.
>> A part of it is the volume of film, it cannot all be verified. So it comes back to trust.
I think the web transaction analogy holds there.
On the one hand, we depend on automated technologies to verify that our data is not intercepted and altered (VPN + encryption).
On the other, we also tend to shop over and over from the same places because we know they're completely fraudulent, which has led to monopoly and monopsony power for some of them and I think we're seeing that in media as well. I recently read that the NYT has more subscribers in California than the LA Times, which is bad, but there it is (even our local library gives me free online access to the NYT, but not to the LA Times or our local paper).
So, for better or worse, we already have Part II, but we don't have a Part I for video.
It strikes me that AI-generated text is the bigger problem. I can't think of any way that can be verified.
And by "bigger problem" I mean from a technology perspective, not a social perspective. From a social perspective, AI video is going to be a massive weapon.
It seems to me most people want to believe what they see. I think I generally mix with intelligent people, who have proved a modicum of intelligence, if not common sense, in their ability to learn.
But the drip drip drip of the extremes seems to be harder to ignore.
If a piece of text has a famous persons name at the bottom, I still generally assume it was written by that person. But somehow a video with a famous person saying those same things no longer can be trusted. I think its hard for us.
Tristan Harris was right in so many ways. We are manipulated by this already.
As one of my favorite quotes goes:
"If a piece of text has a famous persons name at the bottom, I still generally assume it was written by that person."
- Albert Einstein
>> Tristan Harris
I almost mentioned him too. His concept of "human downgrading" helps me understand things better.
Refresher... We used to think that computer intelligence would become a threat to humans at the point that we had so upgraded computers that they could outthink us at every turn.
But what algorithmic media* showed us is that we don't need computer superintelligence, we simply need to deploy algorithms that *downgrade* human intelligence.
So algorithms that encourage anger and outrage and conspiracy thinking and in-group/out-group thinking are capable downgrading humans such that we can be manipulated by relatively dumb actors, either human or computer.
When people who do use algorithmic media say they don't use AI, I always think, "Correct. It uses you."
----------
* I prefer to call it algorithmic media because it's a more accurate description of what it is and what the problem is for two reasons.
1. If we only had non-algorithmic social media (think, The Core), then we really wouldn't have a problem. The problem is not the *social* part, which is good, but the algorithms that promote *anti-social* behavior.
2. Algorithms based on engagement would be less potent if people didn't have platforms to share on, but ultimately even putatively "non social" media has been Buzzfeedified - even the absence of algorithmic Twitter and Facebook and Bluesky, there would still be incentives to produce viral content that would get amplified by Google, for example... sometimes by affiliate marketers ;-)
Quote from: ergophobe on November 08, 2025, 08:52:48 PMsometimes by affiliate marketers ;-)
I don't know what you mean :)
Noah Smith has a recent reflection that is a bit related
"The Internet wants to be fragmented."
https://www.noahpinion.blog/p/the-internet-wants-to-be-fragmented-6b5
QuoteOne thing I notice when I talk to young people these days is that they don't use social media very much — or at least, not the kind that we Millennials used a lot in the 2010s. Zoomers don't want to connect on Twitter, or Facebook, or even Instagram. Instead, they just use phone numbers, like we did in the 2000s.
He is arguing that at the end of the day, people do not actually want the infinite town square model, but want what they had before - an ability to find MY people and connect with them.
He (along with others) see algorithmic media as becoming more and more of a consumption media and less "social".
I like the "Town Square model" idea.
The "town square" idea is Musk's description of Twitter.
https://x.com/elonmusk/status/1507777261654605828?lang=en
Noah Smith is saying that people do not want a giant public town square. In other words, we don't want to be having our intimate conversations from a soapbox. We want to have them around a table in the dining room or while out on a walk together. And we're seeing that in the numbers - the younger generation has gone back to the phone/SMS/WhatsApp as their primary messaging tool.
Most of them do not want followers. They want friends (not "friends" but friends in the way we used that work before 2000.
One of the results is that the "social" platforms are more and more becoming divided into producers (who want a soapbox) and consumers (who like strolling through the park and listening to the crazies on the soapbox). I found that to be true. I still check a couple of exercise physiologists on Twitter and on super rare occasions comment, but I have not posted a thing of my own since 2015 or so as far as I recall.
It's what I like about this place.... No soapboxes :)
Sue has recently got into Instagram. She follows our nieces, the GP500 and F1.
At the same time reading The Dose effect by TJ Power.
The we discuss the very obvious Dopamine hits :)
>>No soapboxes
Sorry. I guess I've been doing it wrong.
Ha Ha, No you do just fine.
We all enjoy your analysis :)
What Rupert said! 8)