Exercises

Every workshop has one core exercise; every async week offers a deep-dive menu at varying time commitments. This page consolidates them for quick reference. The full setup, prompts, and discussion questions live on each weekly module page.

Light ~30 minutes
Standard ~90 minutes
Deep ~3 hours

Stage 1 — Text (Weeks 1–4)

Workshop 1 (May 13) — Build our own ELIZA

Standard Build a Claude Artifact in the spirit of Weizenbaum’s 1966 ELIZA — together. Group exercise. Each person picks a style, tone, and subject of their own — something tied to your discipline, your teaching, or a corner of your life you’d be willing to talk to a chatbot about. The pattern is the constraint; the personality and the topic are yours. Open Claude and prompt it to build a pattern-matching Rogerian-style chatbot Artifact, then layer your style on top. Compare to the masswerk web ELIZA and notice where the ELIZA effect kicks in, sixty years on.

Source: HumanitiesAI/weekone (Three Conversations).

Tour the settings, set your defaults. Walk every panel; set defaults that fit your work; switch models on the same prompt; try voice mode if your account has it. Discord-post what surprised you.
Add a skill or connector. Customize → Add a skill or connector: investigate the menu (skill-creator, document/slide builders, spreadsheet, Google Drive / Gmail / Notion connectors). Try one on a real task. Discord-post what changed.
Upload a document and research. Use Upload file to give Claude your past syllabus, teaching notes, or assignment outline. Ask it to research what others have done in similar classes with generative and agentic AI; iterate. Discord-post what (if anything) Claude surfaced that was useful, and where it felt thin or wrong.
Iterate a poem in an Artifact. Pick a form, iterate ten times, publish as an Artifact. Discord-post the URL and one observation about how the history shaped each draft. Source: HumanitiesAI/weektwo.
Multi-chatbot comparison. Run a complex task (research / poetry / a syllabus paragraph) through two or three of UCF Copilot, ChatGPT free, Gemini free. Notice what each lets you control vs. doesn’t (defaults, refusals, formatting, data handling). Discord-post what you noticed — what’s different in the interfaces and the controls?

Workshop 2 (May 27) — Distant reading + copyright

Standard Build a small corpus, read it across. Bring three to ten texts in your discipline. Upload to a Claude Project. Run the distant-reading sequence: stopword filter → bag-of-words → key phrases → character or theme network → comparative read. End with an Artifact summarizing the patterns. Plus: write one paragraph on the copyright status of your corpus.

Source: HumanitiesAI/weekfour; CriticalMaking2026/exercises/eight_analysis.

Week 4 (async) — Distant reading + Skills

Light One Project Gutenberg book. Pick a single text, run the basic distant-reading prompts in a Project, generate a word cloud and character network. Source: HumanitiesAI/weekfour.
Standard Build your first Claude Skill. Use the skill-creator skill to scaffold a citation-style, alt-text, lecture-note, or syllabus-drafting skill. Save and test on a fresh conversation. Source: HumanitiesAI/weekfifteen.
Deep Design an exercise for your students. Adapt a workflow from W3 or W4 into a 60–90-minute classroom exercise. Write the prompt, scaffolding, deliverable, and rubric.

Stage 2 — Visual and Multimodal (Weeks 5–6)

Workshop 3 (June 10) — Image analysis with Projects + Artifacts

Standard Image-to-text translation set, two tools. Five to ten images you have rights to use. Part A in a Project: alt-text, metadata table, three patterns across the set. Part B in an Artifact: a comparative gallery, grid, or typology. Part C: critique what each tool surfaced and missed.

Source: HumanitiesAI/weeksix; CriticalMaking2026/exercises/four_maps.

Week 6 (async) — Multimodal AI

Light Infographic with Gemini Nano Banana or OpenAI Images 2.0. Take a concept easier to show than say. Generate it in two different tools. Compare what each reaches for first.
Standard Slide deck with Claude Design. Take notes from a lecture, paper abstract, or workshop. Prompt Claude to build a deck via the Design / Artifact slide-deck mode. Iterate three rounds. Compare to building it by hand.
Deep Build your own voice interface in an Artifact. Web Speech API, prompt-driven. Compare to Claude’s built-in voice mode.

Stage 3 — Code Web (Weeks 7–10)

Workshop 4 (June 24) — Agentic AI + GitHub Pages, two paths

Standard Pick a path. Both build and deploy agentically in Claude Code Web — plan → iterate → deploy to GitHub Pages, no local setup. In the live demo, both builds are run in Claude Code Desktop to compare the local workflow with the browser one.

Path A — CV → ePortfolio. Bring a CV / syllabus / project description. Code Web plans, builds a one-page site, deploys to Pages.
Path B — Image set → annotated slideshow. Bring 5–15 images you have rights to. Code Web generates descriptive alt text, renames the files to match their contents, and builds a deployed slideshow. (Adapted from the Image Metadata with Claude Code demo, DHSI 2026.)

Source: HumanitiesAI/weekeleven; CriticalMaking2026/exercises/ten_visualization; DHSI 2026.

Week 8 (async) — Multiple small projects

Pick two or three. Each is small. Each ends in a deployed URL.

Light Polish the W7 site. Two passes: accessibility (alt-text, contrast, headings) and content (any errors).
Standard Static handout → interactive page. Convert a paper handout (glossary, timeline, bibliography) into a one-page interactive site.
Standard Recommender or “find your X” tool. Build a JSON dataset (100 books / 50 films / 30 archives) plus a minimal interactive front end. Source: HumanitiesAI/weeknine.
Standard Public scholarship one-pager. Summarize a piece of your scholarship for a non-specialist audience.
Standard Interactive worksheet for one assignment. Walk a student through one of your assignments in an interactive page with localStorage save.
Light Debug-with-Claude practice. Deliberately break one of the above; fix it only by talking to Claude.

Workshop 5 (July 8) — Course games, planning mode, AI policy

Standard Two halves, both required.

Build with planning mode — a small game or playful tool around a concept from your course. Open with: “Enter plan mode. Do not exit plan mode until I confirm the plan is ready.” Iterate the plan before letting Claude build.
Draft a course AI policy — one page, addressing copyright, attribution, accessibility, equity of access, labor, UDL, and the agentic question. Frame as invitation, not prohibition.

Source: HumanitiesAI/weeknine, HumanitiesAI/weektwelve; CriticalMaking2026/exercises/six_game.

Week 10 (async) — UDL + AI policy, by your discipline

Pick one. All three run in Claude Code (Desktop with Superpowers, or Web).

Light Syllabus review + policy research. Give Claude Code your syllabus; ask it to suggest an AI policy and research existing policies in your field. Compare how Claude frames AI usage versus your discipline’s framing; feed it your draft policy for critique if you have one.
Standard Teaching portfolio on GitHub Pages. Give Claude Code your old and current syllabi, assignments, and other materials you control; ask it to build a job-application showcase that preserves all the content. Host with GitHub Pages.
Deep Course-updating tool via the Webcourses API. Generate an API token in Webcourses, then ask Claude to build a tool that updates your course for aesthetics, accessibility, dates, and broken links. Use Superpowers and planning mode in Claude Code Desktop.

Stage 4 — Agentic Futures (Weeks 11–12)

Workshop 6 (July 22) — Demo + write your `CLAUDE.md`

Light Watch the demo, write a CLAUDE.md. This session is mostly demo (Cowork, Claude CLI, MCP, Superpowers). Your hands-on portion: in a text editor, draft a CLAUDE.md describing your research domain, typical workflows, preferences, and boundaries. No installation required.

Source: DHSI 2026 — dhsi-2026-course-packet (context engineering).

Week 12 (async) — Course proposal / course update

Light Final reflection (500 words). What did you build, what surprised you, what will you carry into fall? Source pattern: HumanitiesAI/finalreflection.
Standard Course proposal / syllabus revision in a Claude Project. Upload syllabus / proposal + your AI policy + one artifact + your CLAUDE.md. Iterate as a critical-reader conversation. Specifically check learning-outcome alignment and draft a 250-word public-facing summary.
Deep One CLI or Cowork session on a personal machine. Optional. Install Claude Code CLI, run one Brainstorm → Spec → Plan → Implementation cycle on a small project (could be your fall course site).

Bring-along checklist (workshop weeks)

Workshop	What to bring
W1 (May 13) — Introducing AI	A laptop. A document, conversation, or short piece of writing for your first prompt.
W3 (May 27) — Textual Analysis	A small corpus (3–10 texts) you have rights to use.
W5 (June 10) — Visual Analysis	An image set (5–10 images you have rights to).
W7 (June 24) — Web Applications	A CV / syllabus or 5–15 images you have rights to. A free GitHub account.
W9 (July 8) — Playful + Policy	A concept from your course. A draft AI policy paragraph. The W7 site URL.
W11 (July 22) — Agentic Futures	All your artifacts (W7, W9, AI policy, Skill from W4). Optional: a personal laptop with Claude Code CLI (or Claude Code Desktop) and Ollama installed, and a Hugging Face account if you want to try fine-tuning.

Exercise Menu