Excerpt:
Read More: https://cointelegraph.com/news/blockcha ... -electionsThe internet is awash with deepfakes, bots and artificial intelligence pretending to be humans — and a host of different projects are racing to come up with an effective solution.
Multiple studies estimate that between 5% and 15% of accounts on X are bots, and Facebook bans hundreds of millions of fake users every quarter.
Online games are also swamped with bots and AIs, used to “grind” through repetitive in-game tasks like mining or farming, or to fake activity for crypto games like Hamster Kombat to receive airdrops.
Sometimes, identifying bots is relatively straightforward, as they can be repetitive, behave erratically or simply make errors a human would almost certainly avoid. But as with toupees, people usually only spot the bad fakes — and the tech is improving so rapidly that it’s becoming difficult to tell what and who is real. Research by the University of Waterloo, Ontario, suggests that people aren’t particularly adept at spotting human replicants.
The March study invited 260 people to sort 20 images of people’s faces: 10 genuine and 10 AI-generated by Stable Diffusion or DALL-E. Only 61% of participants successfully completed the task, lower than the 85% the researchers had projected.
