Photographing the rare brown hyena stalking a diamond mining ghost town | Hacker News Reader

Photographing the rare brown hyena stalking a diamond mining ghost town

bbc.com

63 points165944709122 comments

https://ichef.bbci.co.uk/images/ic/1024xn/p0m8k93b.jpg.webp like something out of Piranesi or Lovecraft

Actually, like one particular scene from the movie "Stalker" by Tarkovsky.

https://images.squarespace-cdn.com/content/v1/58a13eba20099e...

Interesting thing I've noted with wildlife photography competitions is they're often about the backstory of the animal, not so much the photo. Here there photo by itself is not particularly interesting (IMO) but because of the animal and the conditions, it's considered award winning.

"Doing X" vs "doing X in space"

How do we know this isn’t AI?

I don't think these images have this, but there are people working on that problem.

The Coalition for Content Provenance and Authenticity https://c2pa.org/

"The Coalition for Content Provenance and Authenticity, or C2PA, provides an open technical standard for publishers, creators and consumers to establish the origin and edits of digital content. It’s called Content Credentials, and it ensures content complies with standards as the digital ecosystem evolves."

I first heard about this here, which provides a good overview of how t works right now: https://www.tbray.org/ongoing/When/202x/2025/09/18/C2PA-Inve...

Do you have a specific suspicion, or is this more of a general question?

If so, how do we know, you ain't AI?

Well, we cannot know for sure, but we can look at your profile, see your comment history, spot irregularities in your text, etc. - same with the picture. The photographer is a real person, with a history. And as of my knowledge, at very high resolutions, a complete fabrication would take a lot of work, that it could not be spot easily.

Let me put it in another way. How do we know all the pictures in the article and the entire text isn’t fiction?

My question isn’t fact or fiction, it’s just a question.

Yes and I answered that question. It is from the BBC, who has some reputation and the photographer is to be found here https://www.wimvandenheever.com/ who also has a reputation and won awards.

So yes, maybe it is all completely fabricated, but unlikely, as this would destroy his reputation. And as of my knowledge, it is allmost impossible to create high resolution pictures like these, that cannot easily be exposed for fraud (but the tech might have improved, since I last looked into it)

But if you want it way more general, you can also ask Sokrates about, what do we know for sure at all.

https://en.wikipedia.org/wiki/I_know_that_I_know_nothing

You win today. Give it a couple years and you’ll stop winning.

Stop winning what?

Spotting fakes and then not trusting any visual media at all?

Doubt it. Sensors are getting better as well. (More real data, harder to fake).

But I will likely stop debating with internet strangers and rather focus on verified humans, preferably in the real world.

False. More real data doesn’t mean anything when the data is indistinguishable from fake data.

Sensors can’t tell if something is indistinguishable.

Are you aware of the reasons, why accurate weather prediction beyound a short time is not doable?

(Hint, reality is infinitely complex and can only partly modelled, in other words, real world sensor data will always be different, from fake sensor data)

This is not weather. We are modeling a picture or video or text. We are not modeling or simulating reality.

When you try to get the same sensor input to highly sensitive CCD chips, meaning capturing the photons of real world objects, you very much do try to model reality. Otherwise you could not get "photorealistic pictures".

So yes, the tech gets better and better and low resolution stuff can be quite convincing already, so in a few years even quite high resolution(by today standards) might be easy to fake. But even in 20 years, I really, really doubt that the best AI stuff will be close to what the best sensors can deliver.

Follow the trend line. Give it only 5 years until nothing is distinguishable.

You want to bet?

(I think you did not get my point at all)

I get your point. YOu're saying the world needs to be simulated in order to produce identical sensor output.

I disagree. And my evidence for it is that the trendline clearly shows the rate of improvement for pictures is happening at a rate which clearly shows it will produce pictures that will be indistinguishable.

Like this is independent of whether or not the world needs to be simulated. The On the ground evidence literally shows that the projected trendline is that all pictures and all videos from AI will be indistinguishable in 5 years or less. Right now I would say a PORTION of AI and that portion increases every month.

"And my evidence for it is that the trendline clearly shows the rate of improvement for pictures is happening at a rate which clearly shows it will produce pictures that will be indistinguishable."

Are you aware of the advancements in Astronomy for example? Made possible by ever increasing sensor tech.

So ... if you say, AI pictures might get soon to the point, where they are indistingushable from a ordinary mobile phone camera - then I would say maybe, but I really doubt it will be in 5 years.

But when you say sensor data in general will be indistinguishable from AI generated ones, then I disagree.

Good question that probably shouldn’t be downvoted.

A subjective answer is, if you have been there and know this to be real from personal experience.

A more general answer would be, as long as we humans sufficiently interact with reality, we will have a respository of life experience to benchmark against.

Once we cease to do that, and are the product of a life in front of the screen, then we won’t know anymore.

Edit: This place is relatively close to where I live.

"Good question that probably shouldn’t be downvoted."

The very same question like it is, could be literally repeated under any article and is definitely offtopic as it is a general debate how to spot AI and what are the limits of knowledge. Interesting offtopic, so tolerated here if the debate that follows is interesting, but offtopic nevertheless. More ontopic would have been to state why these concrete pictures seem fake.

Give it a couple of years and no one will be able to fully answer the question. I’m quite sure humanity as a whole will in the near future uncover several articles that are like this in quality and citations but find out they are entirely generated and from then on out we won’t know.

We only know now because generated things still have artifacts. That is slowly changing. If the article was written by an AI right now absolutely cannot be fully known.

Reply on news.ycombinator.com