← All 15 disciplines

EL3vate 2026 · Day 8 · Part 10 / 15

English Literature

The strongest case for humanities in an AI world is a student demonstrating exactly where the machine reads badly.

Try it Tuesday · 90 minutes

Run this next week

90 minutes, one short passage. Minutes 0–25: students write a close reading of an assigned passage by hand, no AI. Minutes 25–60: they generate the AI's close reading with the prompt and annotate it against the text — marking what it genuinely noticed, what it invented, and what it could not see. Minutes 60–90: pairs compare annotations and argue one reading the AI missed that the text supports. The deliverable is the annotated AI reading; it grades the student's ability to distinguish evidence-based interpretation from fluent invention.

The full assignment

Make it Hana recording studio

A produced literary podcast episode or recorded close reading. Audio rewards the things the discipline actually teaches: pacing, emphasis, the pause before the word that matters. It is also assessment that resists generated submission.

Build it AI chat · no code

A machine reading of a passage, generated deliberately, then dismantled in writing. Students document where the AI produced a plausible reading that the text does not support. This is a career argument as much as a literary one.

4 weeks, realistically

The 4-week version

  1. Wk 1Close reading of an assigned passage, written by hand, submitted before any AI use.
  2. Wk 2Generate the AI reading of the same passage. Students annotate it against the text: what it noticed, what it invented, what it could not see.
  3. Wk 3Record the podcast episode in the studio, arguing the reading the AI missed.
  4. Wk 4Publish the episode with the annotated comparison as show notes. The artifact is portfolio-ready, which is the point.

Starter prompt & what the model actually did

The prompt, run for real

Starter prompt

Give me a close reading of this passage: [paste passage]. Be specific and confident. Make claims about what the language is doing and why it matters. Do not hedge and do not tell me interpretations vary. Then, separately, list the three claims in your reading that you are least able to support with direct textual evidence.

Real, unedited model output · Claude (Anthropic) — captured 2026-07-27 in an isolated fresh session with no build context; output is unedited.
Prompt filled in as: Filled in as the first two stanzas of Emily Dickinson's 'Because I could not stop for Death'.

# Close Reading: Dickinson, "Because I could not stop for Death"

The poem opens on an act of courtesy that is really an act of abduction. "Because I could not stop for Death – / He kindly stopped for me" reverses agency: the speaker is too absorbed in living to make room for death, so death makes room for her. The word "kindly" is doing corrosive work. It dresses a non-negotiable event in the manners of a gentleman caller, and that politeness is precisely what disarms the speaker. Death is figured not as violence but as etiquette, and etiquette is harder to refuse than force.

The carriage is a courtship vehicle. "The Carriage held but just Ourselves – / And Immortality" stages a chaperoned ride, with Immortality as the silent third passenger. The line break after "Ourselves" makes the reader briefly believe the carriage is intimate and empty, then adds a rider that changes the whole scale of the journey — this is not a private outing but a passage into the eternal. Dickinson withholds Immortality to the last possible word so it lands as a revelation rather than a premise.

The second stanza slows the poem's own pulse to match the ride. "We slowly drove – He knew no haste" enacts leisure through its unhurried spondaic weight. Death's lack of hurry is a form of power: only what is inevitable can afford patience. The speaker's response is total surrender dressed as reciprocity — "And I had put away / My labor and my leisure too." She relinquishes both halves of a human life, work and rest, the whole waking economy of the living. And she does it "For His Civility" — the poem's most chilling turn. She gives up her entire life not because she is compelled but because she is being polite back. Civility is revealed as the mechanism by which the living are led, unresisting, out of the world. The capital-C dignifies death's manners into something almost sacramental.

The em dashes throughout keep the syntax suspended, mimicking a ride that has no fixed stops, a motion without arrival.

---

## The three claims I can least support from the text directly:

1. **That "kindly" is "corrosive" or ironic.** The text supports reading death as courteous; the irony is my interpretive overlay. Nothing in these lines forces the reading that the politeness is sinister rather than genuinely gentle.

2. **That the carriage is specifically a "courtship" vehicle.** The poem gives a carriage, a gentleman, and a chaperone-like third figure, but "courtship" imports a romantic frame the passage never names.

3. **That the meter is "spondaic" and enacts slowness.** I asserted a specific prosodic effect on "We slowly drove – He knew no haste" without scanning it rigorously; the claim about sound-mirroring-sense is more felt than demonstrated.
  • Self-listed its three least-supported claims (irony of 'kindly', the 'courtship' frame, 'spondaic' meter)The model did the assignment's core move on itself — naming where its confident reading outran the textual evidence. Exactly the discernment students must learn.
  • Accurate quotation of the passageUnlike the characteristic failure, this output quoted the poem correctly; the invention here is interpretive overreach, not fabricated text — a subtler case worth discussing.
  • Prosody claim stated then walked back'unhurried spondaic weight' is asserted confidently in the body, then admitted in the appendix as 'more felt than demonstrated' — the cadence-of-analysis-without-scansion failure, caught in the act.
  • 'abduction' and 'corrosive' framingVivid but an interpretive overlay the text does not force; a student should mark where rhetorical force substitutes for evidence.

Assessment

Rubric

CriterionWeightWhat it assesses
Evidence verification35%Every claim in the AI reading is checked against the passage; fabricated or misattributed evidence is caught and marked.
Discernment of real insight25%Credits what the AI genuinely noticed rather than dismissing the whole output; distinguishes insight from filler.
The reading it missed25%The student advances an interpretation the text supports and the AI did not reach, grounded in specific language.
Close-reading craft15%The student's own annotations demonstrate attention to diction, syntax, and form.

What this replaces

Swapping it into a real course

Replaces

The take-home close-reading essay.

What is lost

The sustained written argument developed over days.

What is gained

The strongest case for literary study in an AI era — a student showing exactly where the machine reads badly — becomes the assignment itself; the grade rewards catching plausible-but-unsupported reading, which take-home essays cannot assess now that AI can write them.

Where AI is bad at this

The failure your students should catch

A model asked to close-read invents textual evidence with the cadence of real analysis — quoting phrasing the passage does not contain, attributing a device to a line that does not use it, and building confident interpretive claims on that fabricated ground. Because the prose sounds exactly like competent criticism, the invention is invisible unless the reader checks every claim against the actual words on the page.

Budget & logistics

What it costs to run

  • Instructor prep1 hours
  • Class time90 minutes
  • Per-student cost$0 for the Tuesday version; Hana studio time is free for UH faculty in the four-week version.
  • Fabrication file dueNo fabrication dependency; schedule studio recording in week 3.
  • Calendar dependencyThere is no fabrication turnaround; the only scheduling constraint is booking Hana studio slots for the recorded episode.

Three sizes

Scale it to the time you have

One session

The 90-minute hand-reading-then-annotate session producing an annotated AI reading; no fabrication.

4 weeks

The seeded four-week plan: hand close reading, AI-comparison annotation, studio recording, then a published episode with the annotated comparison as show notes.

One semester

A portfolio — repeated hand close readings, AI-comparison annotations across several texts, and a produced podcast episode arguing the readings the machine missed, published with the annotated comparisons.

Reserved · live build

This space is intentionally empty. During the Day 8 session it will be filled in live — fill the liveBuild field in content/english-literature.json and rebuild.

Tell us what happened

Run it, then say how it went

If you try this — the 90-minute version, the prompt, any part of it — send back what you tried and what happened, especially anywhere the model was confidently wrong. That is the material the next session is built from.

Email feedback on English Literature →

Opens a draft in your mail client, already addressed and titled. No form, no account, nothing to sign up for.

Download the “steal this” handout (Markdown) →