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.
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.