Skip to content

AI Skills for Researchers

AI skills for researchers encompasses the specific competencies economists and social scientists should develop to effectively use AI tools in their work.

Context & Background

The AI skills researchers need are different from those of software engineers or data scientists. Key competencies include:

  • Prompt engineering: Writing effective instructions for AI tools
  • Output evaluation: Critically assessing AI-generated text, code, and analysis
  • Tool selection: Choosing the right AI tool for specific research tasks
  • Workflow design: Structuring multi-step AI-assisted processes
  • Privacy awareness: Understanding data handling implications of different tools
  • Verification methods: Techniques for checking AI output accuracy

Andrew Ng's AI Engineering Skills Map offers a complementary taxonomy from the builder's side — LLM foundations, grounding models with data, building agentic systems, evaluation-driven development, operating in production, and machine learning foundations. Researchers who move from using agents to building them (automated screeners, text-as-data classifiers, replication harnesses) cross into that skill set. Ng's central claim — that a disciplined evals and error-analysis loop is what separates competent builders from great ones — is the builder-side restatement of the verification competency above.

Practical Implications

  • Learn by doing: The best way to develop AI skills is to use AI tools on real research tasks
  • Start with your current work: Apply AI to tasks you already understand well
  • Join communities: Economics AI user groups share practical knowledge
  • Teach others: Teaching AI skills reinforces your own understanding

Key Sources