Episode 01 · 24 August 2026 · 1 hr 10 min

Two Centaurs and Two Human Beings Discuss Progress in Human Geography

Patrick Adler, Ben Gerlofs, Tim Wotherspoon and Giancarlo DeFrancesco on a journal drawing a line, what is lost when the steps are skipped, and whether more research is a good thing at all.

Read the transcript Watch on YouTube

Speakers: Patrick Adler, Geography · Ben Gerlofs, Geography · Tim Wotherspoon, Physics · Giancarlo DeFrancesco, Earth Sciences
Discussed: The Progress in Human Geography statement on AI
Transcript: 1 hr 10 min of conversation, in full. Generated by Zoom, edited by Claude Opus 5.

On a beach, two men argue with two others. Behind one pair a crowd waves placards reading “My Geography is Human, My Reviews are Interpretive, and My AI is Light”. Behind the other, a crowd of cheerful robots. — Generated by ChatGPT Images 2.0 (GPT Image 2)

Contents

What they talk about.

01:29The statement: interpretive reviews, disclosure, and only light use of AI
04:55Editing the Journal of Urban Affairs: you suspect it, and can rarely prove it
06:54Why the frustration is sharper as a peer reviewer than as an editor
08:20Submission agreements, and students bringing generated text to office hours
12:08AI coming for the work worth doing: still washing his own dishes
16:11The hard scientist wants liberation; the social scientist wants better science
17:09Why the review article the journal refuses is the one a scientist would want
19:10The scientific-paper Mad Lib, and where the ideas are supposed to live
20:55Data obfuscation: stripping sample locations before handing over a dataset
28:37Is there a path to teaching discernment? "I don’t, but I want there to be"
30:34ArcMap to ArcPro: the model as friction remover when the tools change
32:32Cory Doctorow, declining quality, and the cancer-screening argument
34:39Monmonier and cartographic critique: rules a model can check, maps it cannot make
37:23Statistical software lock-in, and R-to-Python becoming trivial
41:10Undergraduates meet the technology first, so how can they check anything?
42:49An earnestness to knowledge that is lost when the steps are skipped
43:28Integrity or utility: two readings of what it means to earn a skill
45:32Training checkers instead of experts, and what that concedes
46:49When policing fails you just grade quality, and the essay is always poor
48:35The challenge to the productivity premise: why is more papers good?
49:35Volume as noise, and the cherry-picked citation as a peer review problem
52:26More slop, but also more good papers at the ninetieth percentile
53:59Recent results in mathematics, and the clever mathematician in the loop
54:53Collaborator rather than checker, and why fieldwork resists the handoff
57:07The centaur and the reverse centaur
58:09Why the lab and PI model absorbs AI more comfortably than solo work
59:52If students write worse, should he be teaching writing?
1:03:01Defending your sentences to four classmates who read your draft
1:07:37Four years ago it was bad; now he fears losing the ability to critique

Read the transcript All episodes