Send us Fan Mail
A student turned in a short story this spring that neither of us has stopped thinking about. A man installs an AI system in his home. It does everything for him. Slowly there is nothing left to want, and no one left to talk to. He wrote it as a warning. He is 17.
This started as a workflow episode. Nathan built a college-level writing assignment around Isabelle Hau's "Welcome to the Era of Relational Intelligence" and the full nine-hour Stanford AI+Education Summit, using NotebookLM as the engine and Claude to clean the transcript. We walk through the entire build, step by step, so you can run it in your own room.
Then it became a much larger argument about AI literacy and what school is actually for. We get into cognitive offloading, cognitive outsourcing, and cognitive surrender. We get into active procrastination as a teaching strategy, and why the most creative students are the ones who let an assignment sit. And we get into the dopamine reward system underneath every large language model, the same circuit that drives a honeybee to forage. That is where the bees come in.
One student summed up the whole problem in a single line. AI has a job to do. It cannot not do one. That is the difference between a tool and a relationship, and it may be the most important thing teachers need to understand right now.
Timestamps
00:00 Cold open: the ADHD bee waggle hole
01:35 Why this is a workflow episode, and why Claude is good at cleaning transcript data
02:59 The dataset: the entire Stanford AI+Education Summit, all nine hours
07:34 Bringing the Stanford experience into a high school classroom
09:22 Isabelle Hau and "Welcome to the Era of Relational Intelligence"
12:30 AI and mental health, sycophancy, and what the technology exposes
16:09 The full writing prompt: depict the future, use evidence, propose a turning point
17:27 The build: assembling the notebook and cleaning the transcript with Claude
23:09 A student essay, read in full: the man, the box, and the absence of absence
31:23 The student's breakdown, and Hau on why relational intelligence is indispensable
34:09 The factory model and the danger of siloing the individual
35:00 Sapiens, storytelling, and what set modern humans apart
43:00 Three tiers: cognitive offloading, cognitive outsourcing, and cognitive surrender
44:41 The clearest student line of the year: "AI has a job to do"
46:30 AI literacy as the real work, and the EduProtocols AI Literacy edition
48:33 One screen per table: a setup that beats one-to-one
49:46 Active procrastination as a deliberate teaching strategy
51:16 Adam Grant on why moderate procrastinators are the most creative
52:27 "Where is the work happening?" Nathan does his own assignment, timestamped
59:27 The assignment walked through, step by step
01:00:19 The custom NotebookLM prompt, read aloud
01:11:15 What students built, and the pivot point most of them landed on
01:19:39 The ADHD bees, Huberman Lab, and Dr. Read Montague
01:21:30 The dopamine reward system as the algorithm under every LLM
01:28:31 The first AI-native job, and the gap between the haves and have-nots
01:34:28 Language shapes culture, and AI is shaping language
01:35:21 Adam Aleksic and Algospeak
01:39:12 The Gmail auto-reply story, and engineering a population's language
01:43:44 The inner voice, and what happens when an outside source writes it
01:46:24 Closing on hope, and why this generation gets the last word
Ideas Worth Keeping
Relational intelligence is the counterweight to cognitive surrender. Hau's argument is not a rejection of AI. It is a claim that human connection is the infrastructure everything else depends on, and that infrastructure is now under pressure precisely because AI responds with so little friction. Relational intelligence is under threat and newly indispensable at the same time.
Offloading, outsourcing, surrender. These terms are not codified, so we use them as a working spectrum. Offloading is adaptive and ordinary: a daily briefing pulled from your calendar and email. Outsourcing is genuine collaboration with the machine, and almost nobody has figured out how to do it well yet. Surrender is what the student in this episode dramatized, where a person outsources the need for other people, not just the task in front of them.
Active procrastination is incubation, not avoidance. Open the assignment, do enough to understand what it asks, then let it sit. The thinking continues while you do other things. The best student writing in this unit came from the students who let it cook. Most strong ideas are not the first one you have.
The dopamine reward system is the algorithm. This is the connective tissue of the whole episode. The same circuit that drives a honeybee to forage, and drives some bees to wander off the bee line entirely, is the architecture underneath every large language model. The reward lives in the search more than the finish. Understanding that explains both why AI is compelling and why it cannot replicate a human relationship.
AI has a job to do, and it cannot not do it. Every input is a job. That is its only setting. Human-to-human exchange does not work this way. A person can reject what you say, sit with it, change the subject, or remember something unrelated. An LLM in its current form cannot.
Language shapes thought, and AI now shapes language. If the inner voice is partly engineered by an outside source, identity is implicated. The Gmail auto-reply story is a low-stakes version of a very high-stakes problem.
Recreate the Assignment
The custom prompt loaded into the back end of the notebook, roughly:
"Pretend you have the knowledge and writing style of Isabelle Hau, the organizer and facilitator of this AI and education summit held each year at Stanford. I have added her most recent article, "Relational Intelligence," for content knowledge. Ask respondents questions that push them to question relationality and explore the meaning of this topic as it relates to their futures. Respondents are 16 to 18 years old, attempting college-level work in a freshman composition course. This is paired with a writing assignment that asks students to speculate about our future with artificial intelligence."
Resources
A few of the book links below are Amazon affiliate links. We only link things we actually talk about on the show. If you buy through them, we get a small cut and you pay nothing extra. So, thanks!
Isabelle Hau — "Welcome to the Era of Relational Intelligence"
Stanford Social Innovation Review, Spring 2026
[LINK]
Stanford AI+Education Summit 2026 — All Sessions
Full conference on YouTube, February 11, 2026
[LINK]
Ge Wang — Stanford HAI
Associate Professor of Music, Associate Director at Stanford HAI
[LINK]
McKinsey DELTAs Report — "Defining the Skills Citizens Will Need in the Future World of Work" (2021)
The source for "breaking orthodoxies" as a named, measurable skill
[LINK]
EduProtocols AI Literacy Edition
By Kate Meyer, Nicole Davis, Jon Corippo, and Marlena Hebern
[LINK]
EduProtocols Mindset Episode — WTHTS
[LINK]
Procrastination Episode — WTHTS
[LINK]
Adam Grant — Originals: How Non-Conformists Move the World
The research behind active procrastination and moderate procrastinators as the most creative group
[LINK]
Huberman Lab — "How Dopamine & Serotonin Shape Decisions, Motivation & Learning" with Dr. Read Montague
The episode that connected dopamine reward systems, AI architecture, and bee foraging behavior
[LINK]
Adam Aleksic — Algospeak: How Social Media Is Transforming the Future of Language
On how algorithms shape language, and language shapes thought
[LINK]
Dungeon Crawler Carl — Matt Dinniman
Highly inappropriate. Highly recommended.
[LINK]
Join Jake's Email List (He's sending out the NotebookLM Resource document shortly)
[LINK]
Keep In Touch
Subscribe wherever you get your podcasts. If this one landed, leave a review. It is the fastest way to help another teacher find the show.
The companion newsletter goes deeper on Substack. Free and practical. [LINK] https://whatteachershavetosa
Got a question? We'd love to answer it! Leave us a voicemail on SpeakPipe: https://www.speakpipe.com/whatteachershavetosay
Want more EduProtocols from Jake? Check out his book at Amazon, Barnes and Noble, and more.
