A brief and partial history of generative AI on social media

5000 StyleGAN-generated faces blended together, demonstrating the extremely consistent facial feature placementOne of the more significant developments of the last five or so years on social media has been the escalating use of increasingly sophisticated generative AI technologies for deceptive purposes by spammers, scammers, and political operatives. This article is a very partial chronology of how the misleading uses of these technologies on major platforms have evolved over the last five-ish years, viewed through the filter of my own experience as a social media researcher studying the phenomenon. Most of the examples featured here are from X (formerly Twitter), but similar accounts and networks have turned up on other sites as well.examples of StyleGAN-generated faces used on social mediaThe first form of generative AI to gain traction on mainstream social media platforms was StyleGAN — specifically, the synthetic faces it generates. In February 2019, the ability to create these faces was made easily available to the public in the form of thispersondoesnotexist.com, a website that (at least presently) displays a different GAN-generated face each time it is refreshed. This transformed generative AI from an experimental technology that was largely only available to those with significant computer science skills to a simple tool easily usable by anyone capable of operating a smartphone or web browser. one of the earliest political spam networks to use GAN-generated faces (November 2019)Unsurprisingly, the ability for anyone on the internet to generate large numbers of fake faces was quickly exploited for deceptive purposes. One of the…A brief and partial history of generative AI on social media

Leave a Reply

Your email address will not be published. Required fields are marked *