A.I Created images and video

[TeeDee]

A.I Created images and video

Having seen some rather interesting and some downright disturbing videos produced by AI on SM , I'm actually thinking there maybe a significant problem on the horizon in terms of distinguishing created content from reality.

Anyone feel the same or see it as a novel tool that is as bad as anything else?

 

[Seagull]

I just read through Harari's latest book, Nexus...which has put me in a condition of a sort of concern for the future world and the folk who live in it..
Though, statistically speaking, there's little chance of my being there to see it and, that leaves me believing that those who do grow up and live in that sort of world, will never know anything else...and we are very adaptable.
Together with the stuff that Krauss has on the Tube, the future certainly won't lack for points of interest.

Ceeg

 

[Keith_Beef]

Deepfake audio is just as threatening and probably closer to being a big problem.

A colleague got a call on his mobile phone last week. It was a bot that claimed to represent an electricity company and asked him questions about his accommodation: did he live in a flat or a house, was the flat insulated, did it have double glazing, was the heating gas, electricity, oil, or other... that kid of thing.

He was giving silly answers like he lived in a house with 12 bedrooms, single glazing and no insulation, heated by oil and bills of €20 per month. Oh, that would confuse the analysis, he though.

I don't think that this call was trying to find targets for selling insulation or energy contracts.

I think that it was sampling his voice.

The scammers now have a set of voice sample associated with a phone number and probably his name.

They can then skim social networks to find his connections. Finally they can call his friends, colleagues and especially family, spoofing his number. If the other party picks up, the call will disconnect. If it rings long enough to go to voicemail the scammer's system plays a message fabricated using deepfake with some plausibly (to some people) story about being on holiday in Nigeria, passport and wallet stolen, urgently needing some cash through some service done through grocery shops to tide him over for a couple of weeks and to pay the embassy to issue an emergency passport.

 

[William_Montgomery]

I do think we are creating a bit of a monster with this one. The problems seem pretty obvious to me, any benefits less so.

I saw one example of A.I. offering to write out some form of correspondence on our behalf. You just tap in some basic points you want to put across and A.I. formats the whole thing into a complete, ready to send letter. The idea of sacrificing literacy (arguably one of the most powerful tools at our disposal) for a small convenience, does disturb me quite a lot.

I can't see that we're not going to go down this route, so I think all we can do is keep in mind exactly what the trade off is.

I think I read too many Philip K Dick and Ray Bradbury stories to feel positive about any of this stuff.

 

[shibafu]

I think that it was sampling his voice.

The scammers now have a set of voice sample associated with a phone number and probably his name.

An alarming and very plausible idea.

 

[Chris]

Yeah, it's a real issue. Government officials have been targeted with AI voice and video attempting to gain access to people/information. There was also this case in the US where an employee was conned into transferring $25m by their 'CFO': https://edition.cnn.com/2024/02/04/asia/deepfake-cfo-scam-hong-kong-intl-hnk/index.html

I think we already have a problem on social media with people unable to distinguish fact from fiction, this is going to make it a real epidemic.

 

[ijohnson]

The way people seem to be embracing AI and assistive technology it's like no-one ever watched Terminator or 2001: A Space Odyssey!

 

[Pattree]

I certainly object to Google presenting me with an artificially generated answer to my enquiry.

I just scroll past it and check out the accredited answers that have references that I can follow up.

Of course it won’t be long before the computer can generate references and attributions as fast as I can request them.

Received this example from my son only today.

Meaning of the saying: “Never spread your wolverine on a Sunday”.

 

[homesicksteve]

I was here yesterday - I read a thread about rats.
By the time I got to YouTub there was an AI generated video, just created, on the same subject.
I didn't click.

 

[TeeDee]

Even if we ignore or try to filter where we get our news or media feed with some judicial due diligence , to remove lazy journalism or intent bias from which ever political angle , we fundamentally trust our main sense ( I'd say sight is our main sense ) so much so that its currently very very hard to distrust and discount what one views as long as it falls into plausible parameters.

Make the created visuals toooooo fantastic and we will I think discredit , make the story smaller and more easy to digest , less visual jumps ( example - No giant Cat destroying Tokyo ) - it would be very easier to lead the common citizen to believe whatever they are being presented and influenced.

It must be true... I saw it with my own eyes

 

[Pattree]

Define truth.

(Please )

 

[Wildgoose]

Other side of the coin -

“That wasn’t me caught on camera doing something wrong - must be an AI deepfake”

How many careers have been ended by emerging footage or audio from years back?

 

[William_Montgomery]

Might be quite useful for those with family members that insist on having a 'zoom' chat.

 

[Macaroon]

The very moment I come across anything to do with AI, even voiceovers on Youtube, I'm gone and whatever site or content maker is involved is immediately blacklisted. I think that we, humans, should give a lot of thought to the old maxim - Be very careful what you wish for.

 

[TeeDee]

Define truth.

(Please )

This seems somewhat of a meta philosophical type question - maybe akin to your psuedo-biological/spiritual energy hypothesis about dowsing.

I will however give it a shot - because you asked nicely.

With regards to the topic of AI and swimming lanes of what I'm talking about , regarding visual input - previous and upto this point most peoples main formulation of 'truth/true' has mostly been received ( with regards to news/journalism/world wide reports ) have either been from First hand eye witness events. IE 9/11 " I was there on the ground and saw it with my own eyes " or transmitted by audio visual contents - IE " I watched 9/11 Live whilst the towers were being hit on XYZ news channel."

Of those who say they saw it happening - only a VERY small percentage of those whom witnessed it live were actually in the immediate vicinity - although most of people whom say it happened and witnessed it in real time did so by video.


So I would say upto this point at least our eyes were not super easy to confuse - yes there will be examples and outlier type events that no doubt ' Truth' can be seen from a variety of angles and perspectives and interpreted differently. But it is at least seen - either first hand or via audio visual outlets.

Confusing the human heart and mind ( psyops , bad journalism , propaganda ) is relatively simple -but to confuse or misdirect the eye that the quality and definition of what it is seeing is NOW of such a high quality resolution that it is no longer discernible from reality - because now AI produced visuals are 'as good' as one had come to expect of a simple remote video transmission I think is going to pose a real problem for verification of belief in terms of reality vs digital code.

Thats my attempt - feel like I'm rambling now.

Question back to you - How do you get your news and how do you know its a version of what is actually occurring?

 

[Pattree]

Question back to you - How do you get your news and how do you know its a version of what is actually occurring?

I don’t.

On a spectrum that runs from
Knowledge via Belief to Faith I would suggest that almost the totality of my understanding is belief. I “Know” very little. Hume would say that I can know nothing.

Does E = MC^2? I believe so and predictions based upon it seem to have been useful but I don’t know this. I have read a number of books and articles on the subject and understand its significance to science.*

I have absolutely no time for faith which I define as statements of situations of which there is no evidence anywhere along the communication line. That is, no one, even the author, has had an experience that can be undisputedly linked to the situation described today.

I do not think that “Truth” is always a statement of that which happened or obtained. My truth is an accurate statement of what I believe happened.

The possibility exists that I mistakenly believe that I understand a situation but deliberately distort my description and in so doing accidentally render my description accurate. That is still a lie as far as I’m concerned.

All sorts of circumstances can distort a view of world events so that the most careful of reporters giving the most accurate version of what they believe is happening may be a very long way from the experience of others who themselves cannot know the whole.


It is with a great act of personal sacrifice that I refrain from mentioning another thread or attempt to defend myself against your calumny above

* “Why Does E =MC^2 and Why Should We Care by Brian Cox.
Thoroughly recommended.
In it he makes clear that science does not claim truth.

 

[Falstaff]

Scary future ahead: I've already seen a number of AI generated videos that I've had to take a second look at to spot they are AI. META have stuck a blooming AI search function on my Samsung phone that cannot be removed. It keeps interferring when I want to search for something on my own phone records not the web etc.
The speed and volume of auto-generated AI will eventually outweigh real info, and a search will initially only produce AI junk info and require a lot of digging to get to anything real. As AI gets better at sounding plausible and like us, we will probably lose the ability to tell fact from AI fiction/bad actors.

 

[Pattree]

Nick Bistrom was one of the first to suggest that what we think of as reality is in fact an AI simulation. He put the odds at 50:50.

Musk puts the odds at 99.9% which allows him to claim the moral low ground.

 

[Laurentius]

It is getting worse I am afraid, some of us are now becoming over sensitised to AI images that we reject real pictures when we see them. I was recently caught out on Facebook calling out a picture for being AI because it just did not look right to me, turns out it was genuine

 

[Glow_worm]

The speed and volume of auto-generated AI will eventually outweigh real info, and a search will initially only produce AI junk info and require a lot of digging to get to anything real.

For these reasons there is a genuine possibility the internet could cease to become the useful resource it has been for the last 20-odd years, to the extent that it could actually become easier to regress to using books and libraries as a source of information. Of course, AI has the potential to fill the content of books, but choosing a cut off date of publication would filter it out. Whether publishing companies would want to keep it out of their products is another matter, even if achievable...