Claude Code 101 for Academic Researchers (Bilal)

  • Author/Source: Mushtaq Bilal (academic AI educator), X thread (4M+ views)

  • Original: https://x.com/MushtaqBilalPhD/status/2052338632426467550

  • Key Ideas

  • Frames Claude Code in deliberately non-technical language for humanities/social-science researchers: "you don't need any coding skills"; install it like Zoom or Zotero.
  • Core mental model: instead of bringing your files to an AI app in the browser, you bring the AI into the folder containing all your data.
  • Two reasons it matters for academics: (1) it remembers you, your research, and your writing style across sessions; (2) it can process many files at once — go through 45 PDFs and extract objectives/methodology from each.
  • Concrete starter use cases: qualitative researcher → load interview transcripts, ask for every instance of a phrase, then ask for overarching themes. Quantitative researcher → clean a messy CSV, run descriptives, explain a referee comment.
  • Caveat: "Claude Code is not a replacement for your expert judgment." Academics likely to succeed are those who treat it as a research assistant, not those who outsource thinking and judgment.

  • Setup walkthrough: requires Claude Pro or Max; install from claude.com/download; opens a folder via the GUI; first session writes a CLAUDE.md describing project, research, and writing style.

  • Summary

The first of a five-part introduction aimed squarely at academics who have so far only used browser-based chatbots. Bilal's audience is humanities and qualitative-research faculty, and the framing is engineered to defuse the technical-intimidation barrier: he keeps the word "code" at arm's length, explains the GUI-only desktop app (no terminal), and uses laundry-folding-grade analogies for what Claude Code is doing. The four-million-view reach suggests the framing successfully meets researchers where they are.

The tutorial's substantive contribution is the "bring AI into the folder" mental model — a clean one-sentence explanation of why Claude Code is qualitatively different from ChatGPT-in-a-tab. The introduction of CLAUDE.md as a persistent project instruction file is the other key handoff: it primes researchers for the longer-project organization in Part 2 without overwhelming them in Part 1.

  • Relevance to Economics Research

This is the cleanest non-technical onramp currently public for getting research faculty over the threshold from chatbot-in-a-browser to file-aware agent. Economists supervising RAs or teaching empirical methods can hand it to colleagues who would not read a more technical guide. The qualitative-research framing (interview transcripts, theme extraction) is also a useful counterpoint to the predominantly quantitative orientation of most existing econ-focused Claude Code materials — broadens what counts as a Claude Code use case.