Unmade with AI: AI Detection Remover for Images

Unmade with AI: AI Detection Remover for Images(unmadewithai.com)

2 points by botglen 1 year ago | 6 comments

JohnFen 1 year ago |

This kind of thing is inevitable, of course, but is why the most immediate risk of generative AI is that it casts doubt on the provenance of everything. If we can't tell what's AI generated and what's not, the assumption will have to be that everything is. I think this will only accelerate the degradation of societal trust that is already happening.

botglen 1 year ago | |

The issue addressed here is the grey area of things like Photoshop embedding AI flags if one uses Generative Fill to patch a small gap, extend borders, etc., but the image itself is a real photograph. These flags are carried over to subsidiary works and are hard to remove without wiping normal EXIF data.

localfirst 1 year ago | |

i think you are being overly dramatic. We can already tell if an image is AI generated same reason we can tell if a photorealistic gameplay video is real or not.

Our human brain is incredibly good at spotting contradictions even in the most subtle form.

Low trust societies existed long before AI and its the underlying dynamics in those societies that play a role.

High trust societies will exist regardless of AI because societies are ultimately ruled by law and sense of belonging (identity). Anytime you skimp out on one of those two you enter a negative feedback loop towards low trust society.

JohnFen 1 year ago | | |

> We can already tell if an image is AI generated

Mostly, for now. But if this is some sort of solid truth, then why would a utility like this even be necessary? The only purpose for it (and the one stated in the FAQ) is to make it easier to fool people.