AI Generated Propaganda • Hacker News Board

Re: AI Generated Propaganda

#21

Post by HlessClaudesman » Tue, May 02, 2023, 1:57 PM UTC

Frummy wrote: post #14

Maybe I'm chronically online but I can see instantly if something was AI generated Even had AI generated linkedin profiles message me I guess for scams or hacking or something

Yeah, I would like to credit my powers of AI detection also, but here's the thing: how would you know if you had seen an AI generated piece of media that perfectly mimicked what human could do? To illustrate, the Last Samurai (2003) was a special effects heavy movie but the casual user would barely have noticed any SFX, that was 20 years ago. We are rapidly approaching a similar moment with AI image generation capabilities, if not already past it.

Re: AI Generated Propaganda

#22

Post by morkalork » Tue, May 02, 2023, 1:59 PM UTC

mabbo wrote: post #17

To those saying "Oh I can tell it's fake, obviously!", consider: will your parents/grandparents? And will you still be able to tell in five years, when this tech has had 20 new iterations, each addressing the very tells that right now let you notice it's fake? I know, I know, photoshop and fake pictures have always been around. But now, everyone can do it in 30 seconds. That changes things.

What was life like 150-200 years ago? Back then you had only the written word to go by for news of the outside world and it was just as easily manipulated. I think we've been living through a brief and unique period of truth-y-ness with a plethora of photographic and video evidence.

Re: AI Generated Propaganda

#23

Post by rainsford » Tue, May 02, 2023, 2:04 PM UTC

mabbo wrote: post #17

To those saying "Oh I can tell it's fake, obviously!", consider: will your parents/grandparents? And will you still be able to tell in five years, when this tech has had 20 new iterations, each addressing the very tells that right now let you notice it's fake? I know, I know, photoshop and fake pictures have always been around. But now, everyone can do it in 30 seconds. That changes things.

Does it actually change anything? Maybe AI will led random people make their own pictures, but what's the real impact of that compared with what those people are likely already doing, which is repeating propaganda created by others using only marginally more involved methods like photoshop?

Also the particular image in this tweet doesn't seem like a great example of the power of AI propaganda. Unless I'm missing something, it's a fairly generic image of a nameless person...the propaganda is all in the story attached to it. The same story attached to a stock photo seems like it would have virtually the same impact.

Re: AI Generated Propaganda

#24

Post by skeltoac » Tue, May 02, 2023, 2:05 PM UTC

Off topic. Scrolled a little bit past the thread and almost got caught up in Twitter’s hate bait algorithm. Sexual imagery and I didn’t ask for. This is why I need to remember never to follow links to Twitter, no matter what, ever.

One of the hazards of browsing HN: old darlings like Twitter will be tolerated even after they become NSFW by default.

Re: AI Generated Propaganda

#25

Post by ActorNightly » Tue, May 02, 2023, 2:07 PM UTC

mabbo wrote: post #17

To those saying "Oh I can tell it's fake, obviously!", consider: will your parents/grandparents? And will you still be able to tell in five years, when this tech has had 20 new iterations, each addressing the very tells that right now let you notice it's fake? I know, I know, photoshop and fake pictures have always been around. But now, everyone can do it in 30 seconds. That changes things.

If the Trump era has shown anything, whether the piece of information is real or fake, people will believe and act based on ideals.

Re: AI Generated Propaganda

#26

Post by dukeofdoom » Tue, May 02, 2023, 2:10 PM UTC

AI will bring out the "devil" in our collective consciousness. Midjourney is filled with dark images. People that gravitate to the dark side found an easy way to evoke it and put it out into the world. Political news will follow.

Re: AI Generated Propaganda

#27

Post by wnkrshm » Tue, May 02, 2023, 2:10 PM UTC

mabbo wrote: post #17

To those saying "Oh I can tell it's fake, obviously!", consider: will your parents/grandparents? And will you still be able to tell in five years, when this tech has had 20 new iterations, each addressing the very tells that right now let you notice it's fake? I know, I know, photoshop and fake pictures have always been around. But now, everyone can do it in 30 seconds. That changes things.

Anecdata: Reportedly my great grandma, born in the 19th century viscerally reacted to things that were fictional movies of the 1960s like they actually happened. She could not distinguish between recorded news and recorded drama.

I think it's a matter of our 'priming', what we're used to: If you're used to digital recordings that seem real to actually be real, you'll be tripped up.

If you have no such expectation, you're less inclined to buy into the reality of something that is shown to you. That you have to believe some version of events that you pick up is a false dichotomy.

You can also not believe any version until you get more substantial proof.

Re: AI Generated Propaganda

#28

Post by paganel » Tue, May 02, 2023, 2:13 PM UTC

mabbo wrote: post #17

To those saying "Oh I can tell it's fake, obviously!", consider: will your parents/grandparents? And will you still be able to tell in five years, when this tech has had 20 new iterations, each addressing the very tells that right now let you notice it's fake? I know, I know, photoshop and fake pictures have always been around. But now, everyone can do it in 30 seconds. That changes things.

> To those saying "Oh I can tell it's fake, obviously!",

I couldn't tell the photo was fake from a first or second glance, only after I had started reading the posts and taking a 3rd and 4th look at it I could sort of notice some uncanniness.

But, even then, had no-one told me that the image was fake I would have certainly continue to at least have some doubts about the image's fake-ness.

Re: AI Generated Propaganda

#29

Post by Frummy » Tue, May 02, 2023, 2:13 PM UTC

morkalork wrote: post #22

mabbo wrote: post #17

To those saying "Oh I can tell it's fake, obviously!", consider: will your parents/grandparents? And will you still be able to tell in five years, when this tech has had 20 new iterations, each addressing the very tells that right now let you notice it's fake? I know, I know, photoshop and fake pictures have always been around. But now, everyone can do it in 30 seconds. That changes things.

What was life like 150-200 years ago? Back then you had only the written word to go by for news of the outside world and it was just as easily manipulated. I think we've been living through a brief and unique period of truth-y-ness with a plethora of photographic and video evidence.

Rest in peace to saying "pics or it didn't happen" I guess

Re: AI Generated Propaganda

#30

Post by detrites » Tue, May 02, 2023, 2:14 PM UTC

skeltoac wrote: post #24

Off topic. Scrolled a little bit past the thread and almost got caught up in Twitter’s hate bait algorithm. Sexual imagery and I didn’t ask for. This is why I need to remember never to follow links to Twitter, no matter what, ever. One of the hazards of browsing HN: old darlings like Twitter will be tolerated even after they become NSFW by default.

Swap out the "twitter.com" in the URL for a nitter instance of your choice to see it sans recommendations, ads etc:

https://github.com/zedeus/nitter/wiki/Instances