What happens when AI image models meet social media algorithms?
What is the infrastructure of Slop? How does it get made, and how can we make sense of it?
In this 15-minute video, I walk through the slop pipeline as an interconnected social-technical system, fusing generative AI and the predictive algorithms of social media. I wrote this in March after being invited to contribute to the Slop Salon for a report to be published by Columbia University's School of International and Public Affairs; this is a video version of that presentation.
The insights from that session were also published in a report, "AI Slop and the Information Ecosystem," which Columbia's Institute of Global Politics has made available to anyone, you can find that at the link.
In an effort to better organize and update my website, I've also uploaded the video to the (free) online course offerings, which I am now calling "Thinking Critically About AI Images."
You can click the card below to access the full course talks about the history of AI art, what makes generative AI different, and an overview of some of the issues including data provenance, surveillance, and more. It's a series of videos and transcripts, with guest speakers!
I've also added a short technical introduction to diffusion models based on a workshop I delivered for the metaLab@Harvard's Public AI Summit last summer (no connection to Zuckerberg's Meta).
Hope you enjoy them or make use of them - if you're an instructor or educator please do use these materials in your classrooms if they're helpful! Please let me know if you do.
Where to Find Me
You can find me on bluesky, linkedin or instagram. You can also email me, and I'm always open to invited talks or writing opportunities.