how do we know for sure that the person doesn’t exist just because it’s a computer generated image of a human? What is the probability of the AI generating a human that does exist?
You can verify that by running the code yourself on your computer and comparing the output images with the methods (code) used to generate them.
After a few seconds of loading the page, you'll see some information displayed about the image:
Imagined by a GAN (generative adversarial network)
StyleGAN2 (Dec 2019) - Karras et al. and Nvidia
Don't panic. Learn how it works [1] [2] [3]
Help this AI continue to dream | Contact me
Code for training your own [original] [simple] [light]
Art • Cats • Horses • Chemicals
Another
As for the question of "what is the probability of the AI generating a human that does exist?" ... i'd say the probably is precisely 0.0 since you're looking at pixels and not proteins. ;)
To answer, you'll have to explain what it means for a set of pixels to have a "cellular makeup as [in] a living person". Your question doesn't really follow based on my earlier tounge-in-cheek (but factual) statement about pixels vs. proteins.
Perhaps you're asking about the eventuality of finding a living person that bears a strong resemblance to one of the generated images. That would depend on a totally subjective observation, and you would always be able to take any number of photos of the alleged real subject showing discrepancies with the generated one.
You can verify that by running the code yourself on your computer and comparing the output images with the methods (code) used to generate them.
After a few seconds of loading the page, you'll see some information displayed about the image:
https://www.youtube.com/watch?v=u8qPvzk0AfY https://www.youtube.com/watch?v=dCKbRCUyop8 https://www.youtube.com/watch?v=SWoravHhsUU
https://github.com/NVlabs/stylegan2 https://github.com/lucidrains/stylegan2-pytorch https://github.com/lucidrains/lightweight-gan
As for the question of "what is the probability of the AI generating a human that does exist?" ... i'd say the probably is precisely 0.0 since you're looking at pixels and not proteins. ;)
To answer, you'll have to explain what it means for a set of pixels to have a "cellular makeup as [in] a living person". Your question doesn't really follow based on my earlier tounge-in-cheek (but factual) statement about pixels vs. proteins.
Perhaps you're asking about the eventuality of finding a living person that bears a strong resemblance to one of the generated images. That would depend on a totally subjective observation, and you would always be able to take any number of photos of the alleged real subject showing discrepancies with the generated one.