Another brilliant Java meetup by NY Java SIG (and no, it’s not because I won IntelliJ license raffle). Huge kudos to the organizers for keeping these events coming!
Sean Phillips from Johns Hopkins University presented an astonishing method capable of detecting AI-generated content, be it text (that could be read out), AI-generated audio, or even video.
The topic resonated with me, as I’ve been thinking about methods to detect AI-generated content ever since I started conducting remote interviews earlier this year. Amazingly, it became very difficult to have open-ended conversations, with many responses sounding like readouts. This made me wonder whether that was due to the use of AI tools, or the population of job seekers got so much better.
What’s even more amazing, the method uses AI. Sean’s idea is that AI is based on math, and only mathematical methods can detect mathematically generated content.
The demos were amazing, especially the detection of the deepfake Biden’s calls.
I’m sure there is bright future ahead of this project, will definitely be following the progress and do my best to try it out.
Leave a Reply