Episode 02 · 25 August 2026 · 1 hr 26 min

3 Ps on a Pod: AI Disclosure on Substack and in Academia

Patrick Adler, Parth Goyal and Peter Matanle on disclosure scores, the stigma attached to partial AI use, authorship and credit, and whether transparency survives contact with the incentives around it.

Read the transcript Watch on YouTube

Speakers: Patrick Adler, University of Hong Kong · Parth Goyal, Parth’s Data Stack · Peter Matanle
Discussed: Slop Detective, Parth’s analysis of AI disclosure on Substack
Transcript: 1 hr 26 min of conversation, in full. Generated by Zoom, edited by Claude Opus 5.

Three men talking on a beach beneath a machine labelled “AI Slop Cannon”, which sprays a stream of generated images over them. — Generated by ChatGPT Images 2.0 (GPT Image 2)

Contents

What they talk about.

01:37The Slop Detective study: what Pangram measures, and which categories score highest
05:46Why a fifteen or twenty per cent score carries stigma
09:09The humanities' objections: inauthenticity, plagiarism-by-training, energy and water
15:54The doping analogy: human achievement contest, or problem-solving enterprise?
18:25Nearly everything in academia rests on predecessors who go unsighted
21:29Graduating into a post-AI workforce, and worrying the skills never formed
24:09Authorship and credit, where citations are the only currency academia has
25:44The hackathon: agents making solo work cheaper than collaboration
28:57AI slop as boilerplate with a new engine
33:13Telephone numbers, and what the freed capacity is actually for
34:40Writing a blog to stop the slipping
36:44Picasso at Lascaux: we have learned nothing
40:00Bridging the gap when the lower-order jobs are gone
42:32Fields without benchmarks, and losing a coding moat that was paid for
45:57Returns to expertise: does prior skill make you a better supervisor of agents?
48:51The BS detector, and whether it can be built alongside AI
52:58Handwritten essays plus designed AI tasks, joined by post-task discussion
59:02Tacit knowledge, cookbooks, and why chefs still charge a premium
1:01:27Methodology as the fulcrum of a good study
1:02:55Rigour against enrolment: competition as the thing that drives slop
1:06:22A modest proposal for a doping journal
1:08:31Whether disclosure survives its own incentives
1:10:16Body cameras for academics, mostly tongue-in-cheek
1:14:06Standing by every word of a text that is half generated
1:16:06Disclose when you benefit from perceived ownership
1:16:17Journal editing as a precedent for crediting curation
1:17:37Stigma as gatekeeping, and how enclosure feeds anti-intellectualism
1:21:29Research that can't be explained to an intelligent outsider
1:23:28Learn to Write Badly, and deliberate impenetrability in social science

Read the transcript All episodes