Readings for Tuesday, 14 April
The topic for Tuesday 14 April will be detecting AI-generated content. The required reading is:
- Hanlin Zhang, Benjamin L. Edelman, Danilo Francati, Daniele Venturi, Giuseppe Ateniese, Boaz Barak. Watermarks in the Sand: Impossibility of Strong Watermarking for Generative Models. ICML 2024. [PDF] [Website, which includes a short Video Talk]
In addition to reading the paper, you should try the Pangram AI Detector and read the blog post on How Pangram detects AI-generated content. Your goal is to either generate or find human-written text that it misclassifies as AI generated (a false positive), or generate AI-generated text that it misclassifies as written by a human (a false negative). You can also try GPTZero (but it is too easy to find misclassifications for GPTZero for this to be a challenge).
Optional additional reading:
-
Katherine Thai, Bradley Emi, Elyas Masrour, Mohit Iyyer. EditLens: Quantifying the Extent of AI Editing in Text. ICLR 2026. [PDF]
-
Bradley Emi. AI Conference Papers are Increasingly Being Written by AI: up 370% since 2023. 30 September 2024.
-
Bradley Emi. Pangram Predicts 21% of ICLR Reviews are AI-Generated]. 18 November 2025.
Plan for Thursday, 16 April
I will lead the class on Thursday, 16 April. There will not be any required readings or preparation for the class.
I have a few possible ideas in mind for this, and appreciate any suggestions or requests. I may have an in-class poll on Tuesday to decide what to do in the Thursday class.