Today, an experiment: can Generative AI replace this podcast? There’s a new feature in the GenAI service, Google Notebook LM, that creates an audio discussion about texts you give it. No settings are required. No coding is needed. You only upload a document and give it a few minutes. The output is a conversation in which the uploaded document is discussed. The experiment we’re going to try today is to upload a paper, ask it to generate the conversation, then we’ll talk about the result. Is it any good?
I’m choosing a paper I wrote:
Cain, Joe. 2025. When Fieldwork Goes Wrong, Go Public: George Gaylord Simpson and Anne Roe in Venezuela, 1938-1939. In a book of essays edited by Chris Manias, Palaeontology in Public: Popular Science, Lost Creatures and Deep Time, pp. 221-253, published by UCL Press.
It is available free as an open access book. The abstract is below.
In Part 1, we present the conversation generated by Google Notebook LM. It’s unedited and straight from the processor.
In Part 2, I discuss that conversation with our producer, Capri Huffman. She’s an expert digesting academic work like this. She can spot subtle differences and shifts in emphasis. In one way, I want to treat Google Notebook as just another reader: What does it take away from the paper? What does it claim are the main points, the strengths, and the weaknesses? In another way, I wonder if this approach is a way to identify some of the strengths and weaknesses of the LLM.
This is an episode with two objectives: we talk about my paper (it’s from a project I’ve worked on for a long time and I deeply enjoy the subject), and we talk about a new technology on the scene that wasn’t around last year but now has growing interest and presence on the university landscape.
Abstract for “When Fieldwork Goes Wrong”
In 1938-1939, the vertebrate palaeontologist, George Gaylord Simpson, spent nine months on an expedition through Venezuela with his newlywed wife, Anne Roe, a research psychologist. The expedition was designed as a simple prospecting trip for fossil mammals. But things quickly went wrong, and the couple risked returning with almost nothing to show for the time, effort, and expense. Simpson and Roe showed themselves to be good research entrepreneurs, adapting to their adverse circumstances and ultimately producing a clever range of alternative materials targeted for commercial and amateur communities. This paper explores their recovery of Venezuela and its conversion from technical science to popular adventure. That conversion process was no mere change in register or audience. Simpson and Roe pivoted into popular science and alternative subjects as a deliberate strategy to keep Venezuela working for them in professional and social circles back home.
For a related (boring, academic) presentation I gave in 2008 about Simpson and Roe in Venezuela, see: “Episode 14: Lovers in Lab Coats: When Scientists Collaborate as Husband and Wife | Spring 2008 – Lunch Hour Lectures”
Featuring
Interviewer
-
Capri Huffman, MSc Science, Technology and Society
Interviewees
- Professor Joe Cain
- Google Notebook LLM voices
Host
- Professor Joe Cain, UCL Professor of History and Philosophy of Biology in UCL Department of Science and Technology Studies (STS).
https://ucl.ac.uk/sts/cain
Music credits
Entry and Exit Music
- “Rollin at 5” Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0 License
http://creativecommons.org/licenses/by/4.0/
Music in the Break
- “Sweeter Vermouth” Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0 License
http://creativecommons.org/licenses/by/4.0/
Programme
WeAreSTS is a production of UCL Department of Science and Technology Studies (STS). For more information, visit:
https://ucl.ac.uk/sts/podcast
Visit our episode library:
https://profjoecain.net/we-are-sts/