GANs Unplugged

With the influx of deepfake and style transfer technology in today's news social media, everyone is told that these applications are powered by artificial intelligence deep learning, but too often explanation how it works goes no further. Rather than waiting until second semester computer science grad school to learn about generative adversarial networks (GANs), we propose a classroom activity introduce GANs secondary students. Our Unplugged steps outside traditional environment forms groups students physically act as various components GAN. Students Real Generator (i.e. Fake) both draw sketches animals on index cards, only group has secret, specific instructions their animals. A third Discriminator tries determine if each card receives came from or Fake group. The also makes single feedback mark drawing suggest could be more "real" before passing back drew it. As this game progresses, Discriminators get better at determining real fake, Generators adapt realistic "fake" images. We describe details present our experience running three different summer camps for high