Generative AI for Research

• Course Overview

• Introduction to the Course
• Course Caveats
• AI Content Disclaimer

• Foundations of Generative AI
• History of AI: From Neurons to Neural Networks
• A Lightning Tour of AI (Video)
• But What Is a Neural Network?
• Understanding How Generative AI Works
• An Introduction to Transformers
• Current Generative AI Landscape (May 2026)
• Hands-On Exploration: Testing Generative AI Tools
• Discussion Questions (to prepare before class)

• LLM Architecture Deep Dive
• Training Large Language Models
• Fine-Tuning, RLHF and Alignment
• How AI Image Generation Works
• Discussion Questions (to prepare before class)

• What Does AI Actually Consume?
• Infrastructure, Scale, and the Rebound Problem
• Critical Minerals and AI
• Sustainable AI: What Can Be Done?
• Discussion Questions (to prepare before class)

• Ethical Frameworks and Four Lenses
• Ubuntu, Relational Ethics and the Just AI Framework
• Transparency, Authorship and Integrity
• Applying Ethics: Case Studies and Your Framework
• The Broader Landscape of AI Ethics
• Discussion Questions (to prepare before class)

• The AI Literature Review Landscape
• Free Tools Deep Dive
• Paid Tools and When They Are Worth It
• The Hallucinated Citation Crisis
• Building Your Research Workflow with Claude
• Hands-On Activities and Assessment
• Discussion Questions (to prepare before class)

• Writing as Thinking — Why the Process Matters
• Research Ideation with AI
• AI Writing Tools — Landscape and Honest Assessment
• Scientific Integrity and the Writing Pipeline
• Building Your AI Writing Workflow
• Hands-On Activities and Assessment
• Using AI to Review Your Own Work (Supplementary)
• Discussion Questions (to prepare before class)

• Natural Language to Code — The New Interface
• AI-Assisted Data Analysis in Practice
• Visualization with AI
• Verification of AI-Generated Code
• Building Your Data Analysis Workflow
• Agentic Data Analysis: Giving AI Tools, Not Just Questions
• Hands-On Activities and Assessment
• Discussion Questions (to prepare before class)

• What Multimodal AI Can See, Hear, and Read
• AI and Scientific Images
• Document Intelligence
• Transcription and Audio Analysis
• Video and Multimodal Workflows
• Hands-On Activities and Assessment
• Discussion Questions (to prepare before class)

• The Trajectory of LLM Capabilities
• Three Categories of Failure: Patched, Reduced, Structural
• Where AI Is Now Genuinely Strong
• Illusions of Understanding
• Verification Protocols for a Moving Target
• Hands-On Activities and Assessment
• Discussion Questions (to prepare before class)

• Agents: What They Are and What's New in 2026
• Failure Modes for Multi-Step / Long-Horizon Tasks
• The Current Tool Landscape (Including MCP)
• RAG in 2026
• Advanced Research Tools: A Curated Tour
• Hands-On Activities and Assessment
• Discussion Questions (to prepare before class)

Part A — The Future of AI in Research

• What the Future of AI in Research Might Look Like
• Speculative Futures: A Reading Guide
• The Shifting Research Landscape: Policy, Peer Review, Integrity

Part B — Africa's Sovereign AI Capacity

• Sovereign AI Capacity, and Why Compute Is the Floor
• Data, Languages and African Model-Building
• Policy, Institutions, and Talent
• Discussion Questions (to prepare before class)

• Synthesis and the Integrative Capstone
• Discussion Questions (to prepare before class)

Optional · beyond the free tier · requires a paid Claude subscription and a terminal

Much of this track is adapted, with thanks, from Dominik Lukeš’s course Using AI Agents for Reproducible Research (Oxford e-Research Centre).

Lesson A — Claude Code as a Research Environment

• A.1 — What Claude Code Actually Is
• A.2 — The Honest Case: Cost, Access, and the Disposition
• A.3 — First Contact and the Control Surface

Lesson B — Reproducible Research Workflows

• B.1 — Reproducibility, and the Reproducible Project Folder
• B.2 — Encoding Good Habits: CLAUDE.md, Pre-registration, and Skills
• B.3 — The Reproducible Workflow, End to End