AI for empirical work

Podcasts, videos, tutorials, and research workflows for Seminario de Análisis Empírico at ITAM.

← Seminario de Análisis Empírico

Seminario de Análisis Empírico · Fall 2026

These 54 resources cover using AI, learning, empirical research, writing, and verification. ★ marks eight starting points. The remaining entries are optional; advanced references are labeled.

Use the course’s desktop and RStudio setup for practical work. Technical demonstrations can be watched for their approach. Older resources retain useful examples, while their model, pricing, and interface discussions reflect their publication dates.

Start here

Goal Resource Suggested part
See practical uses 1. Karpathy: How I use LLMs 31:00–1:14:02
Use AI to learn 9. The Mollicks: AI for Students Short video
Study with your sources 11. NotebookLM study guide Select one reading and try a study activity
Work with an agent 14. Panjwani: Codex Desktop 7:25–12:39 and 15:50–47:21
Go from data to a figure 16. Goldsmith-Pinkham: Data Analysis 22:24–35:20
Use AI for writing and thinking 25. Goldsmith-Pinkham: Writing & Thinking 1:27–18:12 and 35:37–43:42
Structure the work 32. Matray: Core Workflow Worked R example
Check the result 33. Matray: Common Failures Failure examples and verification checks

Understanding AI and using it well

  1. ★ Karpathy — How I use LLMs · Video. Practical demonstrations of search, documents, data analysis, and artifacts. Suggested excerpt: 31:00–1:14:02.

  2. Karpathy — Software Is Changing (Again) · Talk. Natural-language programming, partial autonomy, and human–AI collaboration.

  3. Karpathy — Deep Dive into LLMs like ChatGPT · Long video. Training, tokens, hallucinations, and uneven capabilities.

  4. Ethan Mollick — AI’s productivity paradox · TED talk. An accessible introduction with demonstrations.

  5. Mollick — 15 Times to Use AI, and 5 Not to · Essay. Choosing tasks for which you can evaluate the output.

  6. Gábor Békés — Data Analysis with AI: Concepts · Open slides. Concepts, context, applications, and failure modes.

  7. Jascha Sohl-Dickstein — Advice for a Young Investigator in the First and Last Days of the Anthropocene · Talk. Research and career choices under forecasts of rapid AI progress. A related Spanish-language X post discusses this topic; the recording version in that post has not been confirmed.

Learning, reading, and studying

  1. Ethan and Lilach Mollick — Prompting AI · Video. Specifying tasks, supplying context, and iterating.

  2. ★ The Mollicks — AI for Students · Video. Using AI for tutoring and learning while checking its responses.

  3. Wharton — Prompt Library · Reusable examples. Prompts to inspect, adapt, and test.

  4. ★ Google — Six Ways to Use NotebookLM to Master Any Subject · Illustrated guide. Source-based study, quizzes, flashcards, and explanations.

  5. Hannah Fry — Inside NotebookLM · Interview. How a source-based research assistant works.

  6. Hannah Fry and Irina Jurenka — AI in the Classroom · Interview. Learning, engagement, cheating, and dependency. Suggested excerpt: 29:43–39:42.

Data analysis and research demonstrations

  1. ★ Aniket Panjwani — Codex Desktop Course · Video course. Suggested excerpts: 7:25–12:39 and 15:50–47:21, approximately 37 minutes of desktop demonstrations. The economist viewing guide maps the longer course.

  2. Panjwani / VoxDev — AI Agents for Economic Research · Webinar. Research tasks, project organization, reusable instructions, and verification.

  3. ★ Paul Goldsmith-Pinkham — Data Analysis · Demonstration. Census housing data to cleaned data and R figures. 22:24–35:20 focuses on plotting.

  4. Goldsmith-Pinkham — Web Scraping · Technical demonstration. Turning SEC filings into structured data about tariff-risk disclosures.

  5. Goldsmith-Pinkham — Large Datasets · Advanced demonstration. Mortgage data, harmonization, DuckDB, and Parquet.

  6. Békés — Creating Graphs · Worked tutorial. RStudio, descriptive earnings graphs, codebooks, and improving an initial request.

  7. Anton Korinek — Generative AI for Economic Research in Low- and Middle-Income Countries · Webinar and slides. Examples spanning tutoring, writing, coding, and analysis.

  8. Scott Cunningham and Caitlin Myers — The Odd Couple, Episode 1 · Research session. Watch researchers develop a project with AI. The eventual econometrics is advanced.

  9. Cunningham and Myers — Making Maps with Claude Code · Research session. Building and interrogating maps during an actual project.

  10. Development Data Lab’s Tobias Lunt — Generation AI Interview · Podcast. Creating research data about gender norms from historical documents. See also DDL’s project explanation.

  11. Stanford — Empirical Work in the Age of AI · Advanced workshop. Replication demonstrations, verification, research skills, and coordinating agents.

Writing, project instructions, and verification

  1. ★ Goldsmith-Pinkham — Writing & Thinking · Desktop demonstration. Suggested excerpts: 1:27–18:12 for a writing guide and 35:37–43:42 for brainstorming and editing.

  2. Goldsmith-Pinkham — Skills · Demonstration. Creating a reusable paper-summary procedure. Suggested excerpt: 7:33–21:50.

  3. Goldsmith-Pinkham — Integration & Collaboration · Demonstration. Incremental verification and connecting analysis to manuscript outputs.

  4. Claude Blattman — AI Project Folders · Guide. Organizing reusable context and instructions.

  5. Blattman — Teaching AI Your Voice · Guide. Using examples and explicit preferences to improve writing assistance.

  6. Blattman — Stress-Test Any Plan · Browser workflow. Obtaining critical feedback before executing a project.

  7. Blattman — Prompt, Plan, Review, Revise · Guide. Structuring a task, reviewing the approach, and preserving progress.

  8. ★ Adrien Matray — Core Workflow · Worked guide. Planning, reviewing, executing, and verifying an R data-processing task. Part of An AI Guide for Economists.

  9. ★ Matray — Common Failures · Guide. Invented variables, silent sample changes, scope changes, and checks.

  10. Matray — Making It Yours · Optional technical reference. Project instructions, preferences, skills, and settings.

  11. Matray — Skills · Reference collection. Examples for code review, citation checking, documents, and research feedback.

  12. Cunningham and Myers — Human Verification, I Mean, Hannah Verification! · Research session. Independent checking of marriage records before proceeding with analysis.

Templates, collections, and larger workflows

  1. Claude Blattman · Workflow collection. Begin with its Desktop Agent Starter Kit using the course’s existing setup.

  2. Pedro Sant’Anna — Claude Code Academic Workflow · Advanced template. Project instructions, reusable skills, separate reviewers, and verification. Also available as a readable overview.

  3. Antonio Mele — Awesome Econ AI Stuff · Skill collection. Examples for data, figures, writing, and research tasks.

  4. Awesome AI for Economists · Broad directory. Economic data tools, research resources, courses, and talks.

  5. Békés — Data Analysis with AI · Advanced course and instructor reference. Assignments and case studies spanning data processing, text analysis, and econometrics.

  6. Korinek — AI Agents for Economic Research · Advanced reference. Agent concepts and research examples. The opening pages provide an accessible introduction.

  7. Goldsmith-Pinkham — Getting Started · Video. Agents and context. 23:54–29:31 explains the context window.

  8. Goldsmith-Pinkham — Permissions & OpenClaw · Video. 0:00–11:37 introduces permissions. Later implementation material is advanced.

Theory and mathematical reasoning

  1. Ortoleva and Sandomirskiy — Core Uses of AI for Economic Theory · Advanced economics. Model sketching, brainstorming, and checking arguments.

  2. Ortoleva and Sandomirskiy — Can AI Be Creative? · Discussion. Mathematical creativity, attribution, and distinguishing discovery from rediscovery.

  3. Ortoleva and Sandomirskiy — Prompts and Agent Swarms · Advanced workflow. Prompt expansion and separating problem-solving from verification.

Podcasts and broader implications

  1. The Daily — A.I. Is Outsmarting Its Creators · 3 September 2026; 41 minutes. Rogue agents and oversight. Subscription: Apple labels this episode subscribers-only.

  2. The Daily — When A.I. Comes for Your Purpose in Life · 13 October 2026; 33 minutes. Steven Strogatz on mathematics, discovery, and purpose. Subscription: Apple labels this episode subscribers-only.

  3. Panjwani interviews Cunningham — Economics Research and the Returns to Expertise · Interview. Verification and subject expertise. 0:53–9:43 is an accessible opening excerpt.

  4. EconTalk / Tyler Cowen — AI, Employment, and Education · Podcast. An optimistic perspective on work, education, and what students should learn.

  5. Pitchfork Economics / David Autor — AI Won’t Decide the Future of Work—We Will · Podcast. Expertise, distribution, and choices about replacing or complementing workers.

  6. Ideas in Development / Korinek — How to Think Like an Economist About AI · Podcast. Adoption, performance, labor markets, and economic impacts.

  7. Hannah Fry and Iason Gabriel — The Ethics of AI Assistants · Interview. Delegation, anthropomorphism, alignment, and limits on assistance.

Resource notes

Compiled 18 October 2026. Descriptions and suggested excerpts were checked against author or publisher pages. Complete playback and software workflows were not independently tested. Subscription and access requirements can change.

Resources on AI’s future present their speakers’ arguments and forecasts. Check the evidence supporting each claim, just as you would for any other reading.

← Return to the course