Author Topic: AI as a tool for conspiracy videos  (Read 31276 times)

0 Members and 1 Guest are viewing this topic.

Offline pcprogrammer

  • Super Contributor
  • ***
  • Posts: 4670
  • Country: nl
Re: AI as a tool for conspiracy videos
« Reply #25 on: March 08, 2024, 07:22:16 am »
There is no fix for this. Unfortunately people will believe what they want to. The same is true for laws designed to suppress misinformation on social media. It's better to allow people to post it, then people are free to debunk it.

Doesn't work like that, I'm afraid.
What would you do, then?  Allow someone or a panel to judge what is misinformation and what is not?  One such panel was called the inquisition.
Or do you think the inquisition back then was evil and staffed with bad people, and that if you choose better people now, the results will be any different?


Offline pcprogrammer

  • Super Contributor
  • ***
  • Posts: 4670
  • Country: nl
Re: AI as a tool for conspiracy videos
« Reply #26 on: March 08, 2024, 07:36:56 am »
Offence is taken, not given.

Not per se. When one intentionally discriminates someone else it will be both if the other person takes offense of of it.

But that is the problem with this issue, it is hard to describe all the different situations that can arise.

Take for example a woman, you don't know and see for the first time, that looks more like a man, and you address the person as a man. She takes offense, being fed up with constantly being addressed as a man. This is basically her problem. But when you see her again and still address her as a man, than you might do it on purpose and it kind of is your problem.

I have no solution for this either and it looks like society is becoming more sensitive, but history reveals it ain't. Humans have always been this way. Gathering in groups for safety and banish the different minded.

Offline pcprogrammer

  • Super Contributor
  • ***
  • Posts: 4670
  • Country: nl
Re: AI as a tool for conspiracy videos
« Reply #27 on: March 08, 2024, 07:56:49 am »
You can only help people build critical thought and learn how to check facts and evidences through education.
Any attempt to do that for them by law is totalitarian in nature and bound to make them more stupid, more dependent and make the very few that statistically will still exist with an ounce of critical thought become paranoid.

Going to be very difficult with teachers like Soldar described.  :palm:

Offline EEVblog

  • Administrator
  • *****
  • Posts: 39026
  • Country: au
    • EEVblog
Re: AI as a tool for conspiracy videos
« Reply #28 on: March 08, 2024, 08:04:32 am »
I wonder when will we see electronics- based conspiracy videos...
Something like:
"Andrew Miller is a former NASA researcher that built a 1 GHz digital oscilloscope using old CRT television parts and beer cans. He was about to publish his research online, to allow everyone to be able to access top notch electronic instrumentation virtually for free, when a hitman hired by Big Tech teleported his brain on Mars, leaving him in a vegetative state".
Should be the next logical step, after the microwave transformer water pump and the evergreen over unity device.

Next to no market for such content. Which is why all the crap AI content sticks with stuff that gets views, like anything to do with Musk.
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 39026
  • Country: au
    • EEVblog
Re: AI as a tool for conspiracy videos
« Reply #29 on: March 08, 2024, 08:06:27 am »
We're already at that point even without AI interference. Watching the same news on Fox and CNN might leave one uncertain about which media outlet is telling the truth, if any…

Pro Tip: Make your working assumption that all news is fake. ALL of it, and you'll be right a majority of the time.
 
The following users thanked this post: SiliconWizard, Nominal Animal, terminus

Online Zero999

  • Super Contributor
  • ***
  • Posts: 20363
  • Country: gb
  • 0999
Re: AI as a tool for conspiracy videos
« Reply #30 on: March 08, 2024, 08:23:58 am »
Offence is taken, not given.

Not per se. When one intentionally discriminates someone else it will be both if the other person takes offense of of it.

But that is the problem with this issue, it is hard to describe all the different situations that can arise.

Take for example a woman, you don't know and see for the first time, that looks more like a man, and you address the person as a man. She takes offense, being fed up with constantly being addressed as a man. This is basically her problem. But when you see her again and still address her as a man, than you might do it on purpose and it kind of is your problem.

I have no solution for this either and it looks like society is becoming more sensitive, but history reveals it ain't. Humans have always been this way. Gathering in groups for safety and banish the different minded.
That's just bullying and is something everyone should have learned to deal with at school. There should be only a couple of times when it's a problem in adulthood, i.e. the bully is in a position of power, or it's a sustained campaign of harassment and there are already laws which deal with that.  There is no need for extra ones based on protected characteristics.

There is no fix for this. Unfortunately people will believe what they want to. The same is true for laws designed to suppress misinformation on social media. It's better to allow people to post it, then people are free to debunk it.

Doesn't work like that, I'm afraid. A debunker will, by definition, be rational and 'straight-up' whereas the conspiracy nut can and will do things the rational person wouldn't. The conspiracy nut plays to the gallery, and knows how to do so (otherwise they would be down in the noise and a nobody). The straight guy wouldn't dream of stooping to such shit. Thus the conspiracy will nearly always win over the debunking.

And... conspiracies usually contain at least one hard fact, and then use that undisputable fact to justify all the following bollocks. It is really hard to debunk a real fact, so you're on a loser before you even start.
That isn't the case in my experience. I've seen some videos of people putting forward seemingly rational arguments for, what I would say are fringe ideas (I generally avoid the word conspiracy unless it's something truly crazy) and some debunk videos containing good counterpoints, but delivered in a very aggressive manner. It was very apparent the debunked was angry with the person they were rebuking.

I've also seen what happens when videos get deleted for "spreading misinformation". Regardless of whether it really is misinformation, or not, many people are more likely to believe a video if it's deleted because they see it as the authorities censoring content which doesn't suit their narrative.
 

Offline RJSV

  • Super Contributor
  • ***
  • Posts: 2763
  • Country: us
Re: AI as a tool for conspiracy videos
« Reply #31 on: March 08, 2024, 10:21:50 am »
   Reader Vad here, (back a few posts) mentions that, to re-state as I understand, the news on CNN vs on Fox is so different, or of such different bias, that it's hard to reconcile the two, and thus one or the other is less reliable.  True.

   However, like on my smartphone this evening, it's hard to see the news, from both sources when only one of those news outlets gets listed in the first place, at least on Google.
Tonight, following the U.S. presidential speech, the search results completely omit any Fox News reference, whatsoever.
   I see MSNBC, repeated over and over, in the search results.   Also, pressing the 'more video's just revealed similar string of:
   MSNBC, PBS, CBS, videos, each slightly different, and in different order, of course, no Fox News, in spite of a healthy decent large market share.
   That's a non-partisan summary, and fair, that I just did, right ?
   I had typed into the search box,
   'State of the Union'
 

Offline PlainName

  • Super Contributor
  • ***
  • Posts: 7510
  • Country: va
Re: AI as a tool for conspiracy videos
« Reply #32 on: March 08, 2024, 10:46:47 am »
There is no fix for this. Unfortunately people will believe what they want to. The same is true for laws designed to suppress misinformation on social media. It's better to allow people to post it, then people are free to debunk it.

Doesn't work like that, I'm afraid.
What would you do, then?

Don't know. Perhaps, like water running downhill, there isn't a solution.
 

Offline PlainName

  • Super Contributor
  • ***
  • Posts: 7510
  • Country: va
Re: AI as a tool for conspiracy videos
« Reply #33 on: March 08, 2024, 11:01:15 am »
There is no fix for this. Unfortunately people will believe what they want to. The same is true for laws designed to suppress misinformation on social media. It's better to allow people to post it, then people are free to debunk it.

Doesn't work like that, I'm afraid. A debunker will, by definition, be rational and 'straight-up' whereas the conspiracy nut can and will do things the rational person wouldn't. The conspiracy nut plays to the gallery, and knows how to do so (otherwise they would be down in the noise and a nobody). The straight guy wouldn't dream of stooping to such shit. Thus the conspiracy will nearly always win over the debunking.

And... conspiracies usually contain at least one hard fact, and then use that undisputable fact to justify all the following bollocks. It is really hard to debunk a real fact, so you're on a loser before you even start.
That isn't the case in my experience. I've seen some videos of people putting forward seemingly rational arguments for, what I would say are fringe ideas (I generally avoid the word conspiracy unless it's something truly crazy) and some debunk videos containing good counterpoints, but delivered in a very aggressive manner. It was very apparent the debunked was angry with the person they were rebuking.

You angry debunker didn't get anywhere either, I bet, so not being rational and calm isn't a solution. I also bet they just pissed off people so they weren't listening to what they had to say anyway.

T'other problem is that people have a belief in whatever the conspiracy is, and beliefs are mega-hard to undo. Hardly anyone has a belief in non-conspiracy because, well, there isn't anything to believe or not believe in. The alternative to, say, overhead wires causing cancer isn't "overhead wires don't cause cancer" (oh! that's sure to start a conspiracy just saying that!) because there isn't anything. Nothing to debunk. But show that wires have magnetic fields, some indicating tool, try and turn up something bad about magnetic fields... job's a good 'un. So debunking is always going to be a follow-up to those that already believe.
 

Offline SredniTopic starter

  • Frequent Contributor
  • **
  • Posts: 746
  • Country: aq
Re: AI as a tool for conspiracy videos
« Reply #34 on: March 08, 2024, 07:12:56 pm »
Factually speaking, there never was a time when one could trust pictures, videos, or recordings as fact.
Even in the early times of these, a century ago, stuff could be faked, and frequently has.

Well, I wouldn't be so extreme about this.
Yes, pictures - as in photographs - have always been relatively easy to alter. But what about video? Law enforcement has always been partial to security camera footage, and I can hardly recollect a crook that was caught in a crisp video claim the images were faked by AI. Because that ability was beyond most common people, and even most organizations. Before "The mask", only governments could fake videos in believable manner within a limited time span and with limited resources.
Not anymore, tho. Now almost anyone can fake a video were Mr. Honestguy is seen robbing and beating children. Almost no expertise required. There has to be some sort of 'added trust' to either the system that is recording the video (some sort of anti-tampering date-stamped certification that exclude the possibility of further manipulation after the recording - provided it is even possible), or to the subject recording (police, authorities with a legit warrant). The era of the six-fingered hands and messed up background is headed towards its sunset.

The subject of the linked video does not even require faking real people: quite the contrary, they just invent them out of nowhere and automatically create videos that seed false information on a platform that is notorious for the young age of its viewers. Young age that goes hand in hand with an undeveloped sense of critical thinking (for obvious reasons, we have all been through that!). This is recipe for disaster because it can be used to create false premises and prejudices that can be exploited by bad actors.

I believe the damage that can be done goes well beyond 'stealing views' for monetization. It can literally create a generation of useful idiots that will react to determinate situations in a pavlovian way. "Trusting the institution that tells me this product is poisonous? When they buried the research of Dr. Bimbu on the health benefits of hemlock? No way!"
Of course there are more cogent examples that can be made, but they inevitably stray into politics territory, therefore I will avoid mentioning them.

All instruments lie. Usually on the bench.
 

Offline SredniTopic starter

  • Frequent Contributor
  • **
  • Posts: 746
  • Country: aq
Re: AI as a tool for conspiracy videos
« Reply #35 on: March 08, 2024, 07:24:38 pm »
I wonder when will we see electronics- based conspiracy videos...
Something like:
"Andrew Miller is a former NASA researcher that built a 1 GHz digital oscilloscope using old CRT television parts and beer cans. He was about to publish his research online, to allow everyone to be able to access top notch electronic instrumentation virtually for free, when a hitman hired by Big Tech teleported his brain on Mars, leaving him in a vegetative state".
Should be the next logical step, after the microwave transformer water pump and the evergreen over unity device.

Next to no market for such content. Which is why all the crap AI content sticks with stuff that gets views, like anything to do with Musk.

Ok, that was a joke, but what if someone made hundreds of videos showing technicians having trouble with their instruments - for example a blue DMM of the same type you sell, or a red one that reminds of Uni-T, or an orange one that reminds of Agilent, or... just pick a company to damage - without even mentioning the instrument brand.
People getting spurious results, getting shocked by the probes, having the batteries die on them...

It's soon going to be doable without having to hire and pay actors. It probably can be safely done from a legal point of view by avoiding to mention or even show the brand. Brand which can be inferred from other signs, a là Lubitsch. Or, if you live in a country that does not enforce business protection laws, just lie and show the instruments. I am not talking about videos focused on the problematic instruments, but almost legit videos where useful information is actually given but in a large number of occurrences that darn EEVBlog/Uni-T/Agilent/whatever multimeter makes it hard to get to the truth...

Automate the production of these 'tutorials' and disseminate hundreds or thousands of them on multiple channels/platforms and instrument X will be treated like Duracell batteries when it comes to leaks.
It is getting way to easy to sway public opinion...
« Last Edit: March 08, 2024, 07:28:56 pm by Sredni »
All instruments lie. Usually on the bench.
 

Offline Nominal Animal

  • Super Contributor
  • ***
  • Posts: 7198
  • Country: fi
    • My home page and email address
Re: AI as a tool for conspiracy videos
« Reply #36 on: March 08, 2024, 08:14:26 pm »
Yes, pictures - as in photographs - have always been relatively easy to alter. But what about video? Law enforcement has always been partial to security camera footage
They treat security camera footage as they treat eyewitness statements.  It is not like having security camera footage leads to open-and-close case; it's more like having eyewitnesses with no connections to the case provide a clear statement.

Because that ability was beyond most common people, and even most organizations.
There are hundred year old movies with pretty good trick photography and effects.  The reason it hasn't been done (or if it has, not much) is that it has thus far been cheaper to just hire liars with good reputations to provide false eyewitness accounts, with similar effects as security camera footage would have had.

The only real solution is to teach people to think critically for themselves, instead of believing.  Nothing has really changed here, except that lying in a convincing manner is now practically free and available for everyone.
 

Online langwadt

  • Super Contributor
  • ***
  • Posts: 4857
  • Country: dk
Re: AI as a tool for conspiracy videos
« Reply #37 on: March 08, 2024, 08:39:16 pm »
Yes, pictures - as in photographs - have always been relatively easy to alter. But what about video? Law enforcement has always been partial to security camera footage
They treat security camera footage as they treat eyewitness statements.  It is not like having security camera footage leads to open-and-close case; it's more like having eyewitnesses with no connections to the case provide a clear statement.

https://www.bigvalleylaw.com/blog/2021/11/why-eyewitness-testimony-is-notoriously-unreliable/
 

Offline Nominal Animal

  • Super Contributor
  • ***
  • Posts: 7198
  • Country: fi
    • My home page and email address
Re: AI as a tool for conspiracy videos
« Reply #38 on: March 08, 2024, 09:54:25 pm »
Yes, pictures - as in photographs - have always been relatively easy to alter. But what about video? Law enforcement has always been partial to security camera footage
They treat security camera footage as they treat eyewitness statements.  It is not like having security camera footage leads to open-and-close case; it's more like having eyewitnesses with no connections to the case provide a clear statement.
https://www.bigvalleylaw.com/blog/2021/11/why-eyewitness-testimony-is-notoriously-unreliable/
Yep.  This is also why the security camera footage doesn't simply stand on its own, you'll almost always see someone testifying that the person seen in the footage is the person being accused.  In harder cases, you even see experts testifying on whether the footage has signs of tampering or not.  This is like bolstering the eyewitness and their statement: for example, you'd trust an active air force pilot to recognize other airplanes much better than a common person.

Qualitatively, you could say that security camera footage corresponds to testimony by someone trained in observation, or something; but essentially, the same rules apply as to eyewitnesses.  The fact that videos are now much easier to fake, only changes the "default reliability" assigned to security camera footage or cellphone recordings; it does not actually change much in legal proceedings.  It is very similar to if lying became socially acceptable, as if people didn't care about committing perjury anymore.
 

Offline PlainName

  • Super Contributor
  • ***
  • Posts: 7510
  • Country: va
Re: AI as a tool for conspiracy videos
« Reply #39 on: March 08, 2024, 11:04:59 pm »
I used to work for a CCTV company and we would sometimes extract video for the police. Since it comes direct from the recorder and we, the company, have no interest in faking anything, it was treated as reliable.

And that's pretty much the situation with news and stuff. You'd see something on, say, the BBC and have a reasonable trust that it's not been faked. OTOH, the Daily Mail won't have faked stuff per se, but what they do have is 'alternative facts' for it. Similarly, there are Youtube channels that can be pretty sure are straight up and others that you only trust that they are faking something, you just don't know what yet.

I think the problems with the likes of AI video and stuff are not that they fool us, but that they might fool our trusted sources.

That's not to say that we should be 'doing our own research' because, frankly, if you can't trust anything then how can you trust your research? Who are you going to learn from if you can't trust anyone or any source? We offload that kind of thing to those that we are pretty sure can figure it out and explain it to us - our trusted sources.
 

Offline Nominal Animal

  • Super Contributor
  • ***
  • Posts: 7198
  • Country: fi
    • My home page and email address
Re: AI as a tool for conspiracy videos
« Reply #40 on: March 08, 2024, 11:34:46 pm »
And that's pretty much the situation with news and stuff. You'd see something on, say, the BBC and have a reasonable trust that it's not been faked.
And an absolute certainty that the angle and the snippet would match the intent of the reporter/company.

One Finnish newsroom boss publicly admitted that they "do not like publishing good people doing bad things".  If you assume any news organization today is impartial or brutally honest, you're naïve; they are not, and have never been.

I think the problems with the likes of AI video and stuff are not that they fool us, but that they might fool our trusted sources.
Trust, but verify.  Do not "believe".  Build your understanding instead, based on the estimated or assumed probabilities of each assertion being true or false, and be ready to adjust these when you acquire additional data.

I have no "trusted sources".  I only have sources I care about, and a rough understanding of their biases compared to mine.  This is why I can and do read all sorts of "sources" on the net, and even listen to "wonks": I don't need to believe anything, and instead just build a network of associated details, each with a separate level of estimated reliability; there are many "nodes" in there that I believe (= currently estimate) to be complete lies constructed just to harm others.  There is no fact or claim that would "upend" my world, not even finding out we're in a simulation, or that a singular god exists or does not exist.  Everything just adjusts the set of assumptions I base my actions on.

To put it simply, I trust a left-leaning news source to describe events from their perspective relatively honestly, although if any blame is to be assigned, they will prefer to target it on the most suitable right-leaning target.  The exact same applies to right-leaning news sources and left-leaning targets.  The exact same applies if you replace the source and target with any competing pair, be they humans or companies or ideologies or religions.

If you absolutely do need trusted sources, or an authority –– and a lot of people do; their minds are just not wired like mine, and are simply different –– you need to do the comparable amount of work to verify what you trust.  There is no easy answer or free lunch here, and never has been.  Exporting that work to someone else is just giving power to an analog of an inquisition: they will then be in control of your understanding, be that for good or evil.  History just shows that just like many political models, that looks nice and easy on the surface, but in practice tends to lead to suffering and sometimes downright evil, even when instituted only with the purest of good intentions.
« Last Edit: March 08, 2024, 11:38:01 pm by Nominal Animal »
 

Online Zero999

  • Super Contributor
  • ***
  • Posts: 20363
  • Country: gb
  • 0999
Re: AI as a tool for conspiracy videos
« Reply #41 on: March 09, 2024, 10:50:21 pm »
And that's pretty much the situation with news and stuff. You'd see something on, say, the BBC and have a reasonable trust that it's not been faked. OTOH, the Daily Mail won't have faked stuff per se, but what they do have is 'alternative facts' for it.
I don't trust the BBC any more than the Daily Mail nowadays. There have been far too many incidents of them misreporting, twisting and failing to fact check things recently.
 

Offline SiliconWizard

  • Super Contributor
  • ***
  • Posts: 15802
  • Country: fr
Re: AI as a tool for conspiracy videos
« Reply #42 on: March 09, 2024, 11:04:27 pm »
First thing is, not all information you get matters to you. If some information sounds suspicious, but you actually don't need this information, just discard it and move on. It doesn't matter. Only spend time on information that actually matters to you.
I dunno, that may sound very obvious, but will by itself solve 99% of our issues with incorrect information.

We all know that these days, the war on information is striving and is completely polluted by way too many conflicts of interests. Don't try to fight the war by yourself, it's useless. Just try not to be part of a conflict of interest if you can.
Unfortunately, almost anyone who is spreading information as their main (or at least significant) source of income *has* by nature a conflict of interest. Don't *trust*, by default, anyone who's getting paid to say something.

And after you've sorted out the information you really care about, it becomes much easier to check it. Get rid of all the pollution, you won't care if this pollution is true or false.
 

Offline Nominal Animal

  • Super Contributor
  • ***
  • Posts: 7198
  • Country: fi
    • My home page and email address
Re: AI as a tool for conspiracy videos
« Reply #43 on: March 10, 2024, 01:50:31 am »
At some point you'll also realize that whether something is true or not has very little to do with whether people believe in it or are willing to risk the lives of others for.  Thus, follow the money, and look at further than the obvious and immediate results.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf