Overview
This session introduces participants to the art and science of automatic narrative generation — the process of converting structured data and analytical outputs into coherent policy reports, briefings, or storylines using Generative AI tools.
It builds directly on the sentiment and issue analysis work completed in LAB 1, guiding participants through the next stage: transforming data into strategic narratives.
The lab demonstrates how large language models such as ChatGPT, Gemini, and Claude can automatically compose summaries, interpret datasets, and generate multi-section policy narratives in various tones and formats.
Participants will practise prompting, evaluating tone, ensuring factual integrity, and blending visual and textual evidence into human-readable reports.
Learning Objectives
By the end of this module, participants should be able to:
- Explain the concept of data-driven narrative generation and its application in public communication, research, and policy analysis.
- Use Generative AI tools to generate factual, balanced narratives from structured datasets and statistical tables.
- Apply prompt-chaining to control tone, structure, and target audience in AI-generated text.
- Detect bias, inconsistency, or hallucination in automatically written reports and apply corrective prompts.
- Integrate text, tables, and ASCII visualisations into AI-generated policy briefs or executive summaries ready for review or presentation.
Expected Outcome
Participants will be able to produce:
- Fully-automated analytical narratives generated directly from datasets.
- Multiple report versions (executive summary, technical report, media brief).
- Tone-controlled and evidence-based AI outputs suitable for internal or public communication.
- A structured workflow that links data analysis (from LAB 1) with narrative generation (LAB 2).