Spaces:
Running
Intersting but needs work
I have tried a number of images and it is mostly bang on. However it has not correctly identified some 100% human works. As an example, a photo taken of an original watercolour painting reports a large percentage of AI. I was not aware that I came from another planet and was part robot, but I guess I might be.
My advice; treat the results as entertainment. The pictured example is a photo of a handpainted waterecolour on paper, totaly from the minds eye. No filters or effects, just an iPone photo of the pianting.
No classification model is 100% correct, but this doesn't mean that they are 100% wrong (as your "entertainment" comment implies). My suggestion would be simply to not accuse an artist of using AI in a fraudulent manner unless you have tested many example images and they show a consistently high probability of being artificial (e.g. 90%). Even then, I would prefer a more compassionate approach, like asking if they used AI in any way, and if so, how... there's a lot of grey area and room for calm conversation on this topic.
Hi, sorry you have the wrong impression. I think this tool is fantastic. I am not "implying" anything negative.
As stated, I have tried a number of images and it is mostly bang on. I am all for AI. When I use tools like stable diffusion in a project I am upfront about it. But AI is a much broader subject.
Most of us, including me, have been using AI for years, as most of the digital tools we use incorporate some degree of it.
I would love to learn more about how this tool determines its results. I thought being from another planet was a plus :-)