Do you think this kind of internet nuggets will still exist in our soon to be post-AI world ? We won't be able to know who sent a real vs. a fake picture
I guess we will have to rely on extra-net signals: Meta clues from the real world.
For example, the website creator doesn't seem to be looking for profit, nor did they add much oin terms of personal info that would point to him looking for internet clout.
The FAQ page comes across as genuine and, as another commenter put it, whimsical.
It's also all self hosted, and on a unique domain, while mass-content-farmera prefer prefer the zombified audiences of Tiktok and Facebook.
All those signals combine into a high probability of everything on the site being genuine.
Good clues, but what about verifying the authenticity of pictures people send you ? The author here is gathering pictures from others
It'll always be on case by case basis. My mother sending me an awe-inducing picture on WhatsApp? Yeah she probably found it on Facebook, and it's likely it's fake.
In this website's case, I trust the author did enough due diligence to ensure to the best of his abilities that no AI pictures end up on his site. Looking at the submission page (0) he takes submissions by email, and requests the "name of the wildlife sanctuary and the photographer (if known)" which signals he isn't just putting random pictures from the internet.
Text forgeties existed wver since words were written down, and Text has existed for millennia. We had to deal with possible lies and forgeries the entire time.
Photo and Video are very recent inventions, so it was about time they got the same forgery treatment. Now we will have to rely on the same signals of trust as we had before.
Honestly my first reaction to seeing these photos was to wonder if they were AI-generated (I'm not suggesting they are, I just have that response quite often now).
Exact same, my first reaction to the photos were to think they are AI-generated (which amazingly, they aren't).
I actually didn't think that at all, maybe because the opening text was so straight forward, earnest, and pragmatic?
Appearances can be deceiving :\
That in itself is something that AI can leverage, maybe not better-than-average, but way more often, so people have to be on their toes a lot more too. Whether it's images or not.
Interestingly, with images like this they are highly curated for cuteness, clarity, and composition. If nothing else because there are so many photos taken of each owl during the rescue process, across a large number of photo opportunities. So there is often quite a huge variety of material from which to choose one outstanding example for each owl.
This would then make an optimized training set if you wanted to generate realistic facsimiles digitally later on.
When you do the math though, "who" needs a digital facsimile when the vast majority of actual real-world material is far in excess and not being used at all?