Many people complain that, with the rise of AI image software, photography no longer has any veracity. These new AI tools make it relatively easy for anyone to create photo-realistic images of anything.
A photo of Donald Trump shaking hands with Adolf Hitler? Done. A photo of Mitch McConnell in drag? Done. A photo of President Biden picking his nose? Done.
With Presidential elections coming up, we might not believe any photographs of the candidates. We will have no proof that they are real, no indication that they aren't fake. We might no longer believe any photographs we see in any visual medium.
Now, videos can also be faked. The fear of deep fake videos is well upon us. I have seen videos with Arnold Schwarzenegger's face and voice replacing Judy Garland in "The Wizard of Oz" that seem entirely credible. All in good fun, so far.
Any fan of science fiction movies knows that this type of visual trickery has been around for decades. It's also nothing new in terms of photography. Clever retouchers have been able to make all kinds of fake images. Digital tools have made this even easier than ever before. Now, with AI image makers, anything is possible from a simple text prompt.
Unless you look at the hands. So far, AI is doing a terrible job of faking hands. I've seen all sorts of garbled hands: Too many fingers; Hands out of proportion to the rest of the body; Weirdly twisted fingers. It's the easiest tell-tale sign.
But if you're clever in your prompt, you can avoid this problem easily. No doubt, the AI algorithms will eventually figure it all out.
I am here to tell you that photography never had any veracity.
Photographs seem to present reality. They seem to tell the truth. Light is collected through a lens and focused on a sensitive plate of some kind. This "exposure" is somehow turned into a flat picture that seems to represent reality. Sometimes it does. A lot of times it doesn't.
The nature of photography is that it captures something that happens in a split second from a very specific point of view. It shows us something that happened so fast that we probably didn't even notice it. In any case, this "capture" is not necessarily the truth of anything.
For one, a photograph lacks context. We don't know what happened just before or just after that exposure. We can't see everything surrounding the scene that appears to be captured. We don't know what happened just outside of the frame.
Add to that, the flattening of three dimensional space into a flat surface.
You may have taken a picture of someone that seems to have a tree growing out of their head. They didn't. Chance occurrences like this happen all the time with photographs. A well-trained photographer pays attention to everything in the frame. With today's tools, they can remove anything they don't want in the picture. Like your ex-lover., or that cousin you hate.
We have all been fooled into believing that photographs are iron-clad evidence that something happened exactly as shown.
Another problem is how we interpret photographs. Clearly, no one has a tree growing out of their head. Hitler has been dead a long time, so he and Trump never shook hands. Things that seem impossible probably are. But some things in photographs are not so clear. Some times, it's all up to our interpretation.
And when things seem vague, we have a terrible tendency to interpret the worst possible scenarios.
So photographs don't really tell the truth of what happened. Even though they seem to. That is, they don't always tell "the truth".
But what about these AI images? How are they polluting the gene pool of "honest" photography? I'm not sure we can do much about it now. The race is on to create even better fake image-making tools.
One bright spot is that the US Patent Office has rule that AI images cannot be copyrighted. They can be used in any way that anyone wants to use them. That could make it hard to turn these AI robots into profit making tools. In America, if you can't make a profit from it, it doesn't happen.