I brought AI to class today. We'd spent the better part of an hour parsing two academic texts which addressed questions of intersectionality in France, focusing on French Muslims. The research and arguments were congruent but addressed different audiences and debates; one was addressed to political scientists whose analyses of "intersectionality" did not address religion but, the author argued, should. The other was part of a forum defending methodologies like "intersectionality" from critics on the French right who thought them unwelcome North American imports. We had a good time reconstructing the research, the different audiences and contexts, why the two sources were trustworthy - and how one might engage, critique, or go beyond the proffered research.

Then I showed the class a PowerPoint which Notebook LM had generated for me from the two articles. (This took a few minutes, but no specific prompts from me at all.) The slick presentation glommed the two articles together with striking visuals, making a monolithic assertion out of different arguments and interpretations - an aggregated argument neither of our authors had made. The names of the two researchers appeared only once, in tiny print at the bottom of the title slide; their quite different forms of research, reasoning and argumentative contexts - and the openings for engagement and critique - all washed away. 

I didn't spend long on it, though I pointed out various distortions and absurdities in the visualizations. (It gets a lot right, too, of course.) What did I hope the students learned? I told them that Notebook LM is very helpful opening up a single text, but that the work of critical engagement, weighing different sources, was beyond it. That's where you come in! In a week dedicated to the meaning and gift of plural perspectives and experiences, I hope they saw the threat in AI's mindless amalgamation.