Schedule of Readings & Activities

General Info

• Regular bullet points listed under each session indicate what you should READ BEFORE CLASS
• Readings are listed in the order that I think works best
• All readings are available online or through Zotero.
• Most sessions emphasize hands-on activities over lecture

---

1: Mar 23–27

1.1: Introduction and Orientations

No readings yet — these opening questions are meant to surface your prior assumptions before the course starts shaping them. There are no right answers, only more and less examined ones.

• Navigating course materials
• Canvas post on AI keywords:
  • what is INTELLIGENCE?
  • what is THINKING?
  • what is UNDERSTANDING?
  • what is EXPERTISE?
  • Can AI be original?
  • What is the difference between historian history and AI history?

Demonstration: the power of tokens

• What does cilantro taste like? (via https://playground.allenai.org/)

1.2: Intelligence and Machines

• Michael I. Jordan, “Artificial Intelligence—The Revolution Hasn’t Happened Yet”, Harvard Data Science Review 1, no. 1 (2019), 1-8. A computer scientist’s argument that what we call “AI” is mostly statistics and that the real intellectual challenges are about human-machine systems, not autonomous intelligence. Note this is from 2019, which is almost premodern AI.
  • Jordan distinguishes between AI as a marketing term and genuine machine intelligence. Does his skepticism hold up now, or did the revolution happen after he went to press?
  • Jordan is an insider arguing against the hype. Why is that potentially more interesting (or more suspicious) than a humanist making the same argument?
• Ethan Mollick, Co-Intelligence, Chapter 1: Creating Alien Minds, 3–26. Mollick’s opening move: don’t compare AI to human intelligence or to sci-fi robots — treat it as something genuinely new, an “alien mind” with alien capabilities and alien failures.
  • Does the “alien mind” framing help you think more clearly, or does it let Mollick avoid the harder questions about what AI actually is?
  • Compare to Jordan: both are insiders, but they seem to see very different things. What’s driving that difference?
• Why AI isn’t going to make art. This is also in the Zotero library.
  • The argument: AI can reproduce the surface of art — the style, the emotional cues, the structure — without the thing that makes art matter. Is this a real distinction, or nostalgia dressed up as philosophy?
  • How does this connect to Jordan’s “statistics vs. intelligence” argument? Are they making the same point from different angles, or is the art argument doing something different?

Background reading

• Melanie Mitchell, Artificial Intelligence: A Guide for Thinking Humans, Chapters 1–3 (~45 pp.) — a genuinely accessible overview of how AI systems work and why “intelligence” is such a slippery concept.

Class Discussion

• Has the AI revolution actually happened, or are we calling statistics “intelligence” because it sells better?
• Where on the “jagged frontier” does your own expertise sit — and how confident are you that AI can’t already beat you there?
• What makes a good prompt? Is prompting more of a technical skill or a humanistic one?

Information

• Brief overview of how LLMs actually work—tokenization, attention, prediction—just enough to understand that AI generates probabilistically, not through comprehension.

Demonstration — Prompts and models

Prompt engineering is not as much a technical skill as a humanistic one. To write good prompts, you need context, awareness, and empathy. These are fundamental to the intellectual work of the humanities.

2: Mar 30–Apr 3

2.1: What is thinking?

Reading questions are listed under each text to help focus your attention — look at them before you read, not after.

• Alan Turing, Computing Machinery and Intelligence, Mind 59, no. 236 (1950): 433–460. This is as classic as an AI article gets, and we’re skimming/sampling more for flavor than detail. Don’t sweat the technicalities, but focus on what he’s saying about human thinking vs computer thinking.
  • Turing’s first move is to change the question — from “can machines think?” to “can a machine fool a human?” Why does he do this? What does the substitution reveal about the limits of the original question?
  • Turing is writing in 1950 and already anticipating most of the objections people still make today. Which of his rebuttals feel most satisfying? Least?
• Wikipedia: Computing Machinery and Intelligence. This article helps unpack what might be stylistically difficult language. Be familiar with the various ways thinking is defined here and where those definitions come from.
  • Wikipedia catalogs the philosophical objections to Turing’s paper that have accumulated since 1950. Which objections feel strongest to you?
  • The Turing Test is now famous but also routinely dismissed. Has anything replaced it as a useful benchmark for machine intelligence?
• Alva Noë, “Rage Against the Machine”
  • Noë’s claim: computers don’t actually do anything — they’re sophisticated relays, not agents. Thinking requires genuine engagement with a world, which machines don’t have. Convincing, or special pleading for biological exceptionalism?
  • Noë and Turing are essentially in disagreement. What would Turing say in response? Who has the better argument?
• Plato, Phaedrus, excerpts, highlighted in green.
  • Socrates argues that writing creates the appearance of wisdom without the substance — you can read back what you wrote, but the text can’t question you back. How does the dialogue form enact this argument rather than just state it?
  • This is 2400 years old. What has changed about the concern? What hasn’t? (You are, after all, reading it right now, having been assigned it by someone who will discuss it with you.)

Class Discussion: The nature of thinking

• Turing, Noë, and Plato are all, in different ways, trying to protect something they think is genuinely human. Are they protecting the same thing?
• What are the various ways we might define thinking — and which definitions does AI actually satisfy?
• What kinds of thinking are easiest to fake? Hardest? What does that tell us?

Class Discussion: The Dangers of Writing

• What is Phaedrus actually arguing, and how does the dialogue form do work that an op-ed couldn’t?
• Is Plato’s concern about writing just generic anti-tech anxiety, or is there something more specific at stake?
• What would need to be true for the concern to apply to AI — and does it?

2.2: NO CLASS!

3: Apr 6–10

3.1: Technological Resistance

• Calestous Juma, Innovation and Its Enemies: Why People Resist New Technologies, Ch. 1: Gales of Creative Destruction, 11–43. Juma’s core claim: resistance to new technology isn’t irrational — it’s a socially rational response from people who stand to lose something real.
  • What exactly are they losing, according to Juma? Just economic position, or something else?
  • “Creative destruction” is usually told as a story about progress. Juma tells it as a story with victims. How does the same event look different depending on whose perspective you start from?
• Calestous Juma, Innovation and Its Enemies: Why People Resist New Technologies, Ch. 3: Stop the Presses, 68–94. The printing press in the Ottoman Empire was resisted — not by ignorance, but by a specific coalition of interests with real stakes.
  • Who gains and who loses from print in this case? This isn’t just “fear of the new.”
  • How is this case study different from Gutenberg’s Europe? What does the contrast reveal?

Class Discussion: Resistance and Its Reasons

• According to Juma, why do people resist new technologies? Is “it threatens their livelihoods” the whole story, or is something deeper going on?
• How does Juma’s framework help us re-read Phaedrus? Is Socrates a Juma-style technological resister, or is he making a different kind of argument?
• What would Juma say about resistance to AI today? Whose interests does that resistance serve?

Workshop

• Dialogue REMIX pair and share via Canvas
• How can Juma help us think about Phaedrus?

Our collaborative project

• Introduction to course project: Disruptive Expertise

3.2: Getting Dumber

You don’t need to read either of these super closely — they are repetitive enough that you’ll get the idea. But read enough to understand where the authors are coming from. What assumptions are they bringing into their books? Do the arguments make sense, or is it just ranting?

• Neil Postman (1985), Amusing Ourselves to Death, 3–29. Postman argues that television didn’t just change what people watched — it changed how they thought, transforming public discourse into entertainment and making sustained argument culturally impossible.
  • Postman is writing in 1985. Swap “television” for “social media” or “AI.” How much of the argument transfers? How much is period-specific anxiety?
  • Where is Postman at his most convincing? Where does he seem to be mourning a past that may not have existed?
• Tom Nichols (2017), The Death of Expertise: The Campaign against Established Knowledge and Why It Matters, 1–39. Nichols worries that ordinary people have stopped deferring to experts on matters that genuinely require expertise — medicine, climate, economics.
  • Is this always a bad thing? Does it depend on which experts we’re talking about?
  • Compare to Postman: both are worried about the erosion of something serious. Are they worried about the same thing?

Class Discussion

• What is expertise? Is it knowledge, credentials, track record, social trust — or some combination?
• Where is Postman or Nichols most convincing? Where do they shade into ranting rather than arguing?
• How much of either argument applies to AI? Does AI threaten expertise, democratize it, or simulate it?

Prompt Workshop: Creating a new discipline

• General prompting skills are applicable everywhere–let’s experiment!
• Create a new academic discipline and interrogate it
  • What questions do we need to ask in response?
  • How can you see AI’s “jagged frontier”?
    • For more on this, see Mollick’s Centaurs and Cyborgs

4: Apr 13–17

4.1: AI critique

• Safiya Umoja Noble, Algorithms of Oppression, Chapter 1 (“A Society, Searching”) (~30 pp.) Noble documents how Google’s search results produced racially biased outputs — not through malice, but through optimization for profit.
  • What exactly is the mechanism? And does the fact that it’s “not intentional” matter — morally or practically?
  • Is the problem the algorithm, the business model, or the training data? The answer matters for what could actually change.
• Emily Bender et al., “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?”, FAccT ‘21 (2021): 610–623. The paper’s central claim: LLMs generate statistically plausible text without understanding any of it — hence “stochastic parrot.”
  • Is this a meaningful distinction or a philosopher’s quibble? Does it matter whether AI “understands” if the output is useful?
  • The paper also argues that making models bigger creates new risks without clear benefits. Do you find the scale argument compelling?
• Kate Crawford, The Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence (2021), Introduction (~20 pp.). Crawford reframes AI as an extractive industry — with labor costs (underpaid data workers), resource costs (water, rare earth minerals), and environmental costs.
  • Are you comfortable with these costs? Were you aware of them before reading this?
  • Crawford and Noble are both arguing AI is not neutral — but they’re pointing at different mechanisms. What does each emphasize that the other misses?

Class Discussion

• “Algorithms of oppression” — isn’t math neutral? Where does bias enter if not through intent?
• Stochastic parrots: do humans actually understand anything, or are we also sophisticated pattern-matchers? What would Turing say? What would Noë say?
• What present dangers of AI concern you most? Which of the three readings best captures the risk you’d most want addressed?

Demonstration: NotebookLM

Demonstration of Google NotebookLM as a source-grounded research tool.

• Note primary and secondary analysis distinctions!
• What is a trusted source?

Workshop: Start your notebooks!

• Find and upload 2–3 primary sources on technology disruption topic
• Investigate how NotebookLM’s responses are anchored to those specific documents—in contrast to the “anything goes” nature of general chatbots.

With your sources, inquire:

• “What perspectives are missing from this account?”
• “Rewrite this from the perspective of [specific marginalized group]”
• “What sources would I need to find to tell a fuller story?”
• “What do we NOT know about this topic, and why?”

Workshop Debrief

Let’s evaluate AI’s responses. Are the “missing perspectives” it generates substantive or tokenistic? Does it cite real sources? How do you distinguish genuine historiographic insight from AI-generated platitudes about “diverse perspectives”?

4.2: Teach the Teachers

• Ethan Mollick and Lilach Mollick, “Assigning AI: Seven Approaches for Students, with Prompts” (Wharton, 2023). The paper offers seven frameworks for AI-integrated assignments. Skim with a skeptical eye.
  • Do these seven approaches feel genuinely different, or is it one idea about “using AI for learning” in seven costumes?
  • Which approach comes closest to what this course is doing? Which seems most likely to produce students who actually learned something vs. students who know how to look like they did?
• Mollick, Co-Intelligence, Ch. 3: Four rules for co-intelligence, 36–63. Mollick offers four operating principles for working with AI — treat it as capable, push it into your workflow, maintain human judgment at the end.
  • Which rule makes you most uncomfortable? Which do you already follow without thinking?
  • Compare to Noble, Crawford, and Bender from Tuesday: does Mollick adequately account for their critiques, or is he essentially ignoring them?

Background reading

• Bowen and Watson, Teaching with AI, Ch. 11: Feedback, roleplaying and tutors, 209–230.

Class Discussion

• What makes an AI-integrated assignment good vs. gimmicky? What’s the test?
• Which of Mollick’s seven approaches would you actually want to experience as a student?
• Is “the AI is just a tool” a useful framing, or does it obscure something important about what’s happening when you use it?

Assignment review

• Don’t forget to review trusted sources slides
• Check out the AI assignment guide

5: Apr 20–24

5.1: Disruption Assumptions I

• Adrian Johns, The Nature of the Book: Print and Knowledge in the Making, Ch. 1: Introduction, 1–40. Johns demonstrates that the advent of the printing press in Europe didn’t automatically create reliable knowledge, as had been assumed. Trust in printed texts had to be constructed through social and institutional practices. A powerful historical parallel to our current moment with AI!
  • We tend to assume “in print” implies some baseline reliability. Johns shows this was learned, not inherent — it took centuries of legal, social, and institutional work. What does that mean for AI-generated text?
  • What specific mechanisms of trust-building does Johns identify? Who did the work of making print credible, and why?

Class Discussion: Unpacking Truth

• What is Johns actually arguing — and how is it different from the usual “printing press = knowledge revolution” story?
• Trust in printed texts was constructed through specific social and legal mechanisms. Who does that trust-building work for AI, and how is it going so far?
• What would Johns make of AI-generated text? Is it more like early printed books (unverified, pirated, contested) or something genuinely new?
• Critique AI version of “significance of printing press” — what does it get right? What does it smooth over?

Workshop: Peer testing AI assignments

• Trade assignments and try to learn with AI
• Document what works, what’s confusing, where AI helps, where it distracts.

Postmortem

• Each group reports on the assignment they tested.
• What makes an AI-integrated assignment good vs. gimmicky?
• How do you prevent the AI from doing the thinking FOR the student?

5.2: Disruption Assumptions II

• Steven Shapin, A Social History of Truth: Civility and Science in Seventeenth-Century England (University of Chicago Press, 1994), Chapter 1 (“The Great Civility”), pp. 3–41. Shapin shows that in 17th-century England, scientific claims were evaluated partly on the social standing of the person making them — a gentleman’s testimony carried more weight than a craftsman’s, regardless of the evidence.
  • Do you think truth is still this social today? Or have we developed better ways to evaluate claims independently of who’s making them?
  • Compare to Johns: both are arguing that knowledge authority is constructed rather than inherent to the evidence itself. What does this mean for how we evaluate AI-generated claims, which come from an entity with no social standing at all?

Class Discussion: The Social Life of Truth

• Slides overview
• If trust in testimony has always been partly social, what happens when the “speaker” is an AI with no social position, no reputation to lose, and no stakes in being right?
• Shapin and Johns together suggest that our current crisis of AI credibility is not unprecedented — it’s a new version of an old problem. Do you find that reassuring, alarming, or both?

Disruptive expertise workshop

Using NotebookLM loaded with 20 RELEVANT sources (a gathering AND filtering exercise), develop a preliminary historical narrative for your contribution to the collective project. AI helps with synthesis, identifying connections, suggesting additional angles.

Workshop: AI as sounding board

• Ask AI to support the thesis, then to challenge it
• Ask AI for counterevidence, alternative interpretations, and potential objections
• Revise the thesis based on the exchange
• AI is most useful not when it gives you answers, but when it gives you better questions.

Sample prompts

• “What perspectives are missing from this account?”
• “Rewrite this from the perspective of X”
• “What sources would I need to find to tell a fuller story?”
• What historical context is missing but useful?
• “What do we NOT know about this topic, and why?”

Main theme review

• After some work time, everyone gets 1 minute to outline the main themes they intend to cover

Demo: Website setup

• How to set up your own website for your disruptive expertise contribution
• Disruptive Site Setup (start here!)

6: Apr 27–May 1

6.1: Historical Narratives

• Michel-Rolph Trouillot, Silencing the Past, Ch. 1: The Power in the Story, 1–30. Trouillot’s central claim: silences in the historical record are not accidents — they result from choices made at every stage of historical production, from what gets archived to what gets narrated to what gets believed.
  • What are the specific mechanisms of silencing that Trouillot identifies? Where in the process do silences get created?
  • How does AI inherit historical silences? When you ask an AI about an underrepresented community, event, or perspective, what are you likely to get — and why?

Class Discussion: Silences and the Record

• What’s the difference between the past, history, and an AI’s account of either?
• Trouillot, Crawford, and Noble all point at whose knowledge and voices shape the record. Are they making the same argument at different scales, or are the mechanisms meaningfully different?
• If AI learns from the same archives that Trouillot critiques, what is AI “history” actually reproducing?

Lightning Presentations on your disruption

Come prepared to present:

• Brief overview of topic and context
• What’s most interesting about it?
• How is it relevant to AI conversations?
• How did you find sources and evaluate for their utility?
• How did you iterate and direct AI?
• What uncertainties linger?

6.2: Optional Workshop

Come to Mesa Vista Hall 2068 if you need any help or just want to work on your essays!

7: May 4–May 8

7.1: Conclusions

Archival Silences and Structural Authority

• Archival Silences Slides
• Keep in mind the difference between the past, history, and AI outputs.

Peer Review Assignment

• How to do the peer review assignment

Final Course Reflections

• Final Reflection Guide
• Elevate!

AI Keywords Discussion

• What are the common words? Where are the most interesting unique ones?

Expertise + Authority

Drawing on Nichols, Shapin, Johns, and your own semester:

• Can AI be directed to perform as a true expert? What would “true expertise” even mean at this point in the course?
• Does AI-assisted work count as “your” work? What would Shapin say about authorship and credibility?

Education

Drawing on Mollick, Mollick & Mollick, and your own assignment design experience:

• How should academic assessment change in a world where AI can produce competent-looking work on demand?
• When should AI use be disclosed? What’s the argument for always? For it depending on context?
• What should students know about AI when they graduate — and who is responsible for teaching them?

Critical Thinking with AI

Drawing on the full course:

• How can AI help us think more critically, develop skills, and produce higher-quality work? (This was the driving question at the start — what’s your answer now?)
• What has your experience been that you couldn’t have anticipated in week one?
• What’s the most important thing you learned that doesn’t appear in any of the readings?

One last thing

• Concluding Big Points slides

7.2: No Class!

FINALS week