4 ms·
Mildly surprised the quality of this project coming out of MIT. Not only the UI was quite hard to use, I could not get beyond 10 samples. They were annoyingly o
by alpb 5y ago
Mildly surprised the quality of this project coming out of MIT. Not only the UI was quite hard to use, I could not get beyond 10 samples. They were annoyingly obvious. Why would they not choose to not to make this as tricky as possible?
- phs318u 5y agoStrap on your tinfoil hat. It’s clear that state of the art deep fake technology can do better than what’s presented here. What if the whole point of this exercise is to instill (false) confidence in the most people that they have the ability to distinguish a deep fake from the real thing? For the record, I don’t believe this is the case and I fully expect that subsequent iterations of this exercise will be far more difficult.