Resource List

Curated reading for workshop participants who want to go further. Entries are grouped by purpose, not difficulty. Each entry has a one-line note on why it is here.

This list will grow. Suggestions welcome.

Books

Ethan Mollick — Co-Intelligence: Living and Working with AI (Penguin, 2024) The most practical book for knowledge workers adopting AI. Mollick works in academia and writes from direct experimentation — not hype, not panic. Covers how to calibrate trust, what AI does well and badly, and how professional identity changes when AI can do parts of your job. The right book to read before or after this workshop.

Ethan Mollick — One Useful Thing (Substack, ongoing) Mollick's newsletter publishes regular experiments with AI tools in professional and research contexts. One of the best ongoing sources for calibrated, evidence-grounded AI commentary. Free to subscribe: oneusefulthing.substack.com

Reading list

Getting started with AI for research

Teresa Torres — "Claude Code: What It Is and How It's Different" Non-programmer framing of Claude Code, with the "apartment vs hotel room" analogy for why persistent context matters. Good entry point for participants still deciding whether this is for them. → producttalk.org

Every.to — "How to Use Claude Code for Everyday Tasks (No Programming Required)" Practical walkthrough of Claude Code for non-coders. Concrete task examples, realistic expectations. → every.to

Anthropic — Claude Code in Action (free video course) Official hands-on video walkthroughs from Anthropic. Good complement to the workshop for participants who learn better from video. → anthropic.skilljar.com/claude-code-in-action

Anthropic — Claude Academy (free course catalogue) Anthropic's broader official training platform — 25+ structured video courses, free, with quizzes and completion badges. Covers the same ground this book does (Claude 101, Claude Code 101, Claude Cowork, AI capabilities/limitations) plus material the book doesn't: MCP, Subagents, Agent Skills, API/cloud integration. Useful as a structured video companion, or to spot-check this book's technical sections against Anthropic's own current framing. → academy.claude.com/courses

CLAUDE.md and the .claude/ folder

zhaozhiming — "The Complete Guide to Claude Code: CLAUDE.md" Walkthrough of what CLAUDE.md is, how Claude Code loads it at the start of a session, and how to write one well — the single file that gives Claude Code its project-specific instructions and context. → ai.gopubby.com

Avi Chawla — "The Anatomy of the .claude/ Folder" Maps every file inside .claude/ (CLAUDE.md, CLAUDE.local.md, settings.json/settings.local.json, commands/, rules/, skills/, agents/) and what each is for. Useful reference once participants are ready to go beyond a single CLAUDE.md. → levelup.gitconnected.com

Comparing Claude Desktop, Claude Code, and ChatGPT on scoping and memory A saved AI-generated comparison (not authored commentary — flagged as such) contrasting the three tools along two axes: bounded project scopes and memory persistence. Frames Claude Code's distinguishing trait as "environment-centric memory" — state lives in files and git history, not in the model. No external link; kept as a local reference note.

Claude Code in research practice

Chris Blattman — AI workflows for a non-coding political economist Senior policy researcher at U Chicago who describes himself as "never having coded in my life." Documents specific workflows: project dashboards from meeting transcripts, document collection with error-checking, referee report consistency across co-authored papers. The clearest public account of Claude Code used for non-technical research management. → claudeblattman.com

Scott Cunningham — Claude Code series (Causal Inference Substack) Economist documenting his own Claude Code workflow across 31+ installments. One of the most detailed public accounts of a social scientist integrating AI into actual research work. Some posts paywalled. → causalinf.substack.com/s/claude-code

Seth Lazar — "How to use coding agents for philosophy research" Philosopher at ANU built a personal research agent ("Minty") using Claude Code: monitors arXiv and Substack, converts papers to markdown, builds a searchable local knowledge base. Frames it as solving "InfoGlut" — triage-reading at scale so that his actual reading time is spent on pre-filtered material. One of the most intellectually serious accounts of AI as research infrastructure rather than task assistant. → philosophyofcomputing.substack.com

Neuro AI — "Claude Code for Scientists" Practitioner account of using Claude Code in scientific research. Covers file organisation, reproducibility, and the shift from manual scripting to conversational analysis. → neuroai.science/p/claude-code-for-scientists

Benjamin Breen — "Generative AI for Historical Research" Historian at UC Santa Cruz on practical uses of AI in archival and historical work — what it is genuinely good for and where it falls short for humanities research. → resobscura.substack.com

Productivity and the economics of AI adoption

Peter Leyden — "AI Could Trigger the Biggest Productivity Boom Ever" (Big Think, 2026) A knowledge worker's first-person account of 2–3x gains today and a projected 10x within a year or two, used as the springboard for a macro argument: AI could produce the largest productivity boom in history, but only if wealth redistribution policy catches up. Contains the best short version of the compounding-productivity arithmetic (0.6% vs. 1.9% vs. 4% annually over 25 years). Usefully optimistic but structurally honest. → bigthink.com

AI in humanities and social science

Tom Pepinsky — "Agentic AI and Social Science Research Practice" Political scientist (Cornell) on how agentic AI changes social science. The distinction between execution and interpretation — what AI can do and what requires the researcher — is the clearest short statement of the core tension in this workshop. → tompepinsky.com

Messing & Tucker — "The Train Has Left the Station" (Brookings) Policy-oriented view of agentic AI in social science: productivity cases alongside systemic concerns. Useful for researchers thinking about field-level consequences, not just personal workflow. → brookings.edu

Christopher Pollin / DHCraft — LLM DH Summer School Digital humanities curriculum for working with large language models in humanities research. CC BY 4.0 — freely reusable materials, good for course integration. → chpollin.github.io/llmdh

Christopher Pollin / DHCraft — "Promptotyping with Claude 4" Humanities scholar using Claude Code to prototype digital scholarship tools without formal programming training. The Stefan Zweig example is a concrete model for this workshop's audience. → dhcraft.org

David Berry / DHNow — "Vibe Decoding: Building the Critical Code Studies Workbench" Digital humanities scholar building a research workbench with Claude Code as a non-programmer. Honest account of what worked and what needed iteration. → digitalhumanitiesnow.org

Critical perspectives

Andy Hall / Niskanen Center — "Can AI 'Vibe Research' Replace Social Science?" Political scientist making the case that AI shortcuts in research produce outputs that look like findings but lack the epistemic work that makes them meaningful. Required reading alongside enthusiasm about efficiency gains. → niskanencenter.org

Christopher Pollin / DHCraft — "Vibe Coding" Critique of vibe coding from a digital humanities perspective — what goes wrong when the tool does the work and the researcher loses track of what was done. → dhcraft.org

Harvard Shorenstein Center — "New Sources of Inaccuracy: A Conceptual Framework for Studying AI Hallucinations" Conceptual framework for thinking about hallucination as a class of problems, not a single failure. Useful for grounding the risk discussion in something more precise than "AI makes things up." → misinforeview.hks.harvard.edu

AI in qualitative methods

Xule Lin — "Interpretive Orchestration" Claude Code infrastructure for qualitative research built around an "epistemic partnership" model: four specialised agents, a three-stage methodology (solo practice → human-AI collaboration → scholarly synthesis). Rejects both naive automation and dismissal — argues AI can offer scale and interpretive depth together, but only if the researcher maintains interpretive authority at each stage. → github.com/linxule/interpretive-orchestration

Child Trends — "Case Study: Using AI to Analyze Qualitative Interview Transcripts" Research nonprofit account of AI-assisted analysis of sensitive interview data (sexual/reproductive health). Documents the full pipeline: de-identification, two-stage coding, batch sizing to reduce errors, mandatory human review. Key finding: AI initially produced overlapping themes; iterative human correction was essential. Honest about limits and conditions of use. → childtrends.org

Xu (2026) — "Doing Thematic Analysis in the Age of Generative AI" (Qualitative Inquiry) Peer-reviewed treatment of what AI assistance means for thematic analysis — when it is methodologically defensible and what disclosure obligations it creates. → journals.sagepub.com

Ozuem et al. (2025) — "Thematic Analysis in an Artificial Intelligence-Driven Context" (Sage) Companion piece: practical and ethical dimensions of AI-assisted thematic analysis in social science. → journals.sagepub.com

Linguistic Variation (2024) — LLM annotation of Early Modern English texts (De Gruyter) Peer-reviewed study of Claude for pragmatic annotation of historical corpora. Finding: inter-coder agreement between Claude and humans is comparable to disagreement between human coders on ambiguous cases. The operationally relevant result — AI is as good as a second human coder for well-defined annotation tasks, without requiring programming expertise. → degruyterbrill.com

AI as research collaborator

Gottweis et al. (2025) — "Towards an AI Co-Scientist" (Google DeepMind, arXiv:2502.18864) Google DeepMind's paper on AI as a hypothesis-generating, literature-synthesising research collaborator. Not a tool review — a conceptual proposal for what it means to work with AI on research problems rather than use it as a utility. → arxiv.org/pdf/2502.18864

Related