5 ms·
If you develop technology that can reliably detect deepfakes (i.e., fn is_deepfake(m: media) -> bool), then couldn't an AI be trained to defeat it?
by marshray 4y ago
If you develop technology that can reliably detect deepfakes (i.e., fn is_deepfake(m: media) -> bool), then couldn't an AI be trained to defeat it?
- f1shy 4y agoI think it would lead to arms race.
- waste_monk 4y agoYes... but you can then train a model to detect the improved fakes. And then train a better deepfake model, and so on, forever. This technique of pitting two AI against each other (a generator and a detector/discriminator) is called a generative adversarial network, and it's used a lot for unsupervised training.