Report 2605

OpenAI just launched the "AI Text Classifier" to identify texts generated by AI. Tried it, and IT DOES NOT WORK. https://platform.openai.com/ai-text-classifier

Using my Python ML book published in 2015:

1. @randal_olson's foreword: unclear
2. my preface: possibly AI
3. paragraph from Ch1: likely AI

I mean, this is a funny example, but I already feel bad for students who might get penalized for their essays in the future because of this.

If you deploy a model like this, pls share a confusioon matrix.

Models like this can cause real-world harm due to educators adopting this for grading. So let's add some transparency about False Positives and False Negatives.

First page from Shakespeare's Macbeth. What!?

Hypothesis: can't feed anything created before 2022, because it was all training data.