Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI-generated 'poverty porn' fake images being used by aid agencies

theguardian.com

Post reply on HN
Print view View original

71–80 of 222 posts

Re: AI-generated 'poverty porn' fake images being used by aid agencies

#71

Post by zdragnar »

stinos wrote: post #55
SkyeCA wrote: post #28

Earlier quoted context omitted.

> How can you possibly make due diligence when everyone around you is incentivized to lie? You can't and the correct response is a total lack of trust by default because that's the easiest way to protect yourself. They really are out to get you(r money).

> You can't and the correct response is a total lack of trust by default because that's the easiest way to protect yourself. Rather off-topic, but it's funny how this principle applies in the exact same way when it comes to traffic for instance. It is unfortunate it has to be like that, but not trusting any other traffic and assuming they can at any point do the thing you'd least expect them to do is just safer. Espe…

There's nothing unfortunate about it, because you don't have perfect information. There's plenty of ways for traffic to behave unexpectedly for reasons that are out of their control (deer running into road, tire blows out, etc).

This is why driving schools teach defensive driving. You can't control anyone else but yourself.

Re: AI-generated 'poverty porn' fake images being used by aid agencies

#72

Post by AppleBananaPie »

carlosjobim wrote: post #48
robin_reala wrote: post #11

The reasonable solution to this is to legally require every single image and video (or at least those used in commercial / charitable / public sector / journalistic settings) to be marked with a flag indicating the level of amendment. If it’s untouched apart from cropping / colour grading etc no mark needed. If it’s had some touch-up work require that to be flagged at minimum in the image metadata. And if large chunk…

Unfortunately for you, "the government" is not an almighty and all-seeing god, even though hackers worship the government in such a way. Thus those regulations and solutions you are calling for are simply impossible and it won't matter how much money and prayers you send to the government. It won't change reality.

I'm probably misunderstanding because this seems bad faith.

Isn't it obvious the government isn't an all seeing god?

All regulations are imperfectly written and imperfectly enforced but many seem to have impact.

I'm not sure if regulating this is feasible or not but it's not immediately obvious it's impossible to some extent.

Re: AI-generated 'poverty porn' fake images being used by aid agencies

#73

Post by jnovek »

SkyeCA wrote: post #28
A4ET8a8uTh0_v2 wrote: post #6

I was thinking about it, because wife was telling me story from work, where a woman was scammed with AI generated stuff and her colleague was a little too nonchalant about it ( 'it is on her to do her due diligence' ). And it made me annoyed. How can you possibly make due diligence when everyone around you is incentivized to lie? We do have a concept of fraud, but advertising seems to be able to move around its edges…

> How can you possibly make due diligence when everyone around you is incentivized to lie? You can't and the correct response is a total lack of trust by default because that's the easiest way to protect yourself. They really are out to get you(r money).

I wouldn’t describe it as the correct response, more like “the best one we can think of right now.”

Living this way is that it’s really exhausting, saps quite a lot of joy out of life and makes people more lonely. It’s far from optimal and a more sustainable option would be to work our way back to the community trust we used to enjoy just ~30 years ago.

Re: AI-generated 'poverty porn' fake images being used by aid agencies

#74

Post by joegibbs »

robin_reala wrote: post #11

The reasonable solution to this is to legally require every single image and video (or at least those used in commercial / charitable / public sector / journalistic settings) to be marked with a flag indicating the level of amendment. If it’s untouched apart from cropping / colour grading etc no mark needed. If it’s had some touch-up work require that to be flagged at minimum in the image metadata. And if large chunk…

How does that help though, if a charity uses an entirely raw unedited photo that is still totally fabricated or misleading?

I think that’s the real problem - it’s not really any better if a charity uses a paid actor on a set, or finds the absolute poorest person in the worst street in the country and uses their photos as if they’re representative of an entire country for the purpose of soliciting donations. If the AI pictures were actually representative of the situation then I wouldn’t think it’s all that bad.

Re: AI-generated 'poverty porn' fake images being used by aid agencies

#75

Post by zdragnar »

at-fates-hands wrote: post #50
SkyeCA wrote: post #28

Earlier quoted context omitted.

> How can you possibly make due diligence when everyone around you is incentivized to lie? You can't and the correct response is a total lack of trust by default because that's the easiest way to protect yourself. They really are out to get you(r money).

>> You can't and the correct response is a total lack of trust by default because that's the easiest way to protect yourself. This has slowly eaten away at the idea that we used to live in a high trust society that has now completely transformed into a society where you cannot trust anything, ever, in any capacity. I felt like I had kind gained back some control from not clicking on any links in emails and using my p…

Communities used to be smaller, even in big cities, so the number of scammers you encountered was fewer.

Now that long distance calls are free, the Internet connects you to everyone else on it, and paperwork is digitalized across entire populations, the reach - and therefore statistical likelihood of encountering- scammers is much higher.

Re: AI-generated 'poverty porn' fake images being used by aid agencies

#76

Post by myth_drannon »

jncfhnb wrote: post #25

I have suspected that I have seen some of these for Palestine in recent months. Not saying this has any implication on the broader reporting of it but… shit we’re really kind of screwed if media starts portraying everything via fictitious dramatization imagery. Especially on social media.

Yes, it's called Pallywood.

Re: AI-generated 'poverty porn' fake images being used by aid agencies

#77

Post by Igrom »

resoluteteeth wrote: post #13

One thing that's interesting to me is that a lot of the ai-generated images shown in the screenshot of the stock image site seem like they are emulating the style of national geographic photographer Steve McCurry who has had some controversy due to editing his images and also allegedly has staged images (which is something that you aren't supposed to do for documentary photography), and also I think has been criticiz…

Well said. I have always tried to understand Baudrillard's process of formation of the simulacrum. Only it's a pity that we are no longer content with reimagining mint chocolate or pumpkin spice, now it's time to stereotypize real destitution and sorrow.

Anyway, who will want to continue to fork out money over fraudulent images of victims? How can I know that this or that agency represents actual people in need?

In Zurich, an Oktoberfest-related festival (Zuri-Wiesn) held inside a large tent-like structure put up within the main train station's main hall has recently concluded. On the outside, on both sides, the tent was adorned with large-format prints of vaguely Germanic rural folk in celebratory garments holding up their pint. All generated! Blurry facial features, malformed metal buttons. I'd argue that already damaged the joviality and earnestness of the event. And that's just beer...

Re: AI-generated 'poverty porn' fake images being used by aid agencies

#78

Post by jampekka »

myth_drannon wrote: post #76
jncfhnb wrote: post #25

I have suspected that I have seen some of these for Palestine in recent months. Not saying this has any implication on the broader reporting of it but… shit we’re really kind of screwed if media starts portraying everything via fictitious dramatization imagery. Especially on social media.

Yes, it's called Pallywood.

[flagged]

Re: AI-generated 'poverty porn' fake images being used by aid agencies

#79

Post by pessimizer »

lotsofpulp wrote: post #52
jstanley wrote: post #40

Earlier quoted context omitted.

It has failed you in ways you can't see, because you've driven away the most reliable people by showing them upfront that you don't trust them. I assume everyone is telling the truth until they've given me a reason not to assume that, and I'd say that hasn't failed me yet.

This is too simplistic. There are varying levels of trust for myriad interactions. You can trust someone to drive properly and not hit you. You can trust someone to not rob you as you walk by them. You can simultaneously not trust them enough to give them money if they ask for it without sufficient proof. Or not trust them enough to invite them into your home.

I think there's a problem with the way the word "trust" is used altogether. I trust certain people to drive badly, I trust them to forget appointments, and I trust them not to offer to pay the bill.

"Trust" being something defined as both exclusively positive and a virtue in the person that gives it is just meant to soften up marks, and is probably instilled into us by marketing and advertising. It's certainly not taught by any religion, who concentrate on being worthy of trust, not trusting.

You trust people because you know them well, and you know what they will do when confronted by certain situations because you have seen them react to similar situations before. Before you know them well, you're just operating from stereotypes and comparisons with other people you've known who have similar mannerisms. Or you're projecting your own thought processes into their heads. A "level" of trust is just a confidence level in a prediction.

If you don't have a friend who has had to sacrifice something for you, you don't know if you have any friends at all. That's the real reason it's hard to make friends as an adult. But after you've been through stuff with people, you'd be a fool not to know they'll come through in this new, far-less grave circumstance.

I honestly think the problem is that we're trying to make people fungible in a weird secular market religion, and they're not. Trust is not a credit score.

Re: AI-generated 'poverty porn' fake images being used by aid agencies

#80

Post by jnovek »

pessimizer wrote: post #58
A4ET8a8uTh0_v2 wrote: post #31

Earlier quoted context omitted.

Sigh. People will adjust. My point, and a subtle one at that, was that it does not bode well for the society of low trust ahead.

What people say to respond to this critique is that high-trust people were always a problem. Low-trust societies have procedures in place that nobody ever considers just skipping because we're both good guys and you're from Boston, too, right? High-trust societies are self-regulating scammers paradises. Filled with people citing Occam's Razor, and affinity fraud. Even worse, for minorities the procedures never get dr…

The kind of high-trust we need to get back (at least here in the states) is the kind between neighbors on the same block. Of course it’s never been wise to trust someone just because you’re both from Boston, but in the past neighbors generally trusted each other and socialized.

It’s particularly striking when I visit my parents’ house. I’m a millennial and I don’t even know my neighbors names and my parents, who are boomers, routinely talk, text and visit their neighbors. Their neighborhood (at least among the old people) feels like a community. Mine does not.

Post reply on HN