Exploring the Future of AI-Generated Videos and Deepfakes

Video: https://www.tiktok.com/@distilledscience/video/7514164723143789855

Transcript

Can someone please explain to me what the real-world use case is for AI-generated video and deepfakes? Oh, hey! I'll give you four, and explain some tech that'll make deepfakes less of an issue in the future. One, democratizing creativity. In the last century, movies were made by big studios with big budgets. Execs decide to make unnecessary sequels and reboots because it's a safe bet that'll make money, recouping costs. In a few years, I bet we'll see a single creative teenager from an underprivileged home win an Oscar. Two, and this is the one I'm super excited for. Education and health. Kids would care a lot more about history if they could see it, or speak with dead people. So many topics across science and medicine would be so much easier to learn with really good visual aids, not just textbook illustrations. Imagine a magic school bus episode for every topic. Three, accessibility and reach. The Malaria Must Die campaign partnered with David Beckham to create this video, a petition to end malaria, with him being deep faked to speak natively in nine languages. And thus resonate much better than my METFA And thus resonate much better with native audiences. Selfishly, why should a non-native English speaker have to listen to my videos in their second language when instead they could see a perfect AI version of me talking in their native tongue? For historical preservation, This museum in Illinois used AI to let visitors speak with recreations of Holocaust survivors. It felt like Fritzie was right there on the stage. Yes, deepfakes are scary. In the next few years, we're going to see them some terrible things before the right protective systems are put in place. But making them illegal is not the answer. Legislation rarely wins the arms race against new technology. But here's where the future is headed. In the last century, we operated under a paradigm of assume true until proven fake. But from now on, when you see a video or image, you're going to have to assume fake until proven true. The future will be defined by something called Providence tracking. There's a coalition of all the big tech companies working on this called C2PA. It works almost like a blockchain embedded in an image or video. by the camera, then it's added to by any editing software like Photoshop, so that when a news organization or social media viewers sees the final product, they can click on that little icon and see the full history of where that image came from and how it was edited. All fully verified by some complicated math and cryptography. But it'll take some time for this to get implemented everywhere, and until then, we're going to all have serious trust issues.

Additional notes

Replying to @Avisha - 🧬Science Made Simple what do you guys think? Do we need legislation, and if so, what? #ai #aivideo #deepfakes

References

  • Malaria Must Die campaign with David Beckham mentioned in transcript; direct URL not listed in workbook.
  • Illinois museum AI Holocaust-survivor recreation example mentioned in transcript; direct URL not listed in workbook.
  • C2PA provenance tracking coalition mentioned in transcript; direct URL not listed in workbook.