Topics
Both lists are candidate topics. The number of groups per class depends on class size, so the final list may differ. The topics can also help you prepare for the profile-specific knowledge test that is held twice a semester. For more background on your profile, see the AI Engineer and Creative Technologist profile pages.
Each topic can only be chosen once per class. See First week: choose your topic for how to choose.
AI Engineer
These topics come from the semester's self-study material, which is available for free online:
- PDSH: Python Data Science Handbook by Jake VanderPlas
- DLWP: Deep Learning with Python (third edition) by François Chollet and Matthew Watson
The foundational knowledge page of your profile has more on the self-study material.
| Topic | What it covers |
|---|---|
| Data wrangling, EDA & visualisation | NumPy, pandas and Matplotlib as the basis for exploring and preparing a dataset. PDSH: ch. 1, ch. 2, ch. 3, ch. 4 |
| Classical machine learning | Decision trees and random forests, linear regression, support vector machines, naive Bayes, and unsupervised methods such as clustering, PCA and manifold learning. PDSH: ch. 5 |
| Foundations & fundamentals of deep learning | What deep learning is, the mathematical building blocks, and practical fundamentals: overfitting, generalisation and validation. DLWP: ch. 1, ch. 2, ch. 4, ch. 5 |
| Computer vision: image classification | Convolutional neural networks applied to image classification. DLWP: ch. 8 |
| Time series forecasting | Forecasting techniques for sequential and time-based data. DLWP: ch. 13 |
| Language models & text classification | Representing and classifying text with neural networks, how language models work, and the Transformer architecture behind current LLMs. DLWP: ch. 14, ch. 15 |
Creative Technologist
These are technical AI themes, not generic design methods. Problem analysis, research through design, evidence-based decisions, prototyping and user testing still matter, but you use them as ways of working within every theme. See the knowledge page of your profile for the foundations behind them.
The external resources are recommendations
The links marked Recommended are optional suggestions to help you get started. You are not required to use them. Use them, or other resources you find, in whatever way helps you and your group.
| Topic | What it covers |
|---|---|
| Local language models & human–AI interaction | Small local models, prompting, local processing, quality, privacy, conversational interaction and user control. Recommended: Ollama quickstart, Guidelines for Human-AI Interaction |
| AI sees, hears & interprets | Image, sound, pose or hand recognition; training data, labels, error behaviour, context and exclusion. Recommended: ml5.js beginner's guide (The Coding Train), Gender Shades |
| Trust, uncertainty & transparency in AI interactions | How interfaces present AI output, limitations, sources, uncertainty, correction and control. Recommended: Explainability + Trust, Errors + Graceful Failure (Google People + AI Guidebook) |
| Autonomy & human control in AI systems | AI that advises, performs steps or uses tools; consent, boundaries, oversight and recovery after errors. Recommended: Feedback + Control (Google People + AI Guidebook), Building Trustworthy Agents |
| Data, classification & meaning | How systems categorise, score, select or recommend; the assumptions involved and ways to make them visible. Recommended: Datasets Have Worldviews (Google PAIR), Excavating AI |
| Generative AI as creative & critical material | Local text generation and, where feasible, multimodal interaction; visualisation and the influence of generated output on making, choices and authorship. Recommended: Machine Learning for Artists, Introduction to Generative AI and LLMs |
The recommended technology baseline for Creative Technologists is:
- HTML, CSS and JavaScript for the frontend (React may be used where it helps). Recommended: Learn web development (MDN), The Modern JavaScript Tutorial, and for React the React quick start
- Ollama with a small local language model on your own laptop, as the first choice for LLM-based artefacts. Recommended: the Ollama quickstart and the Ollama JavaScript library
- ml5.js for in-browser machine-learning experiments. Recommended: A Beginner's Guide to Machine Learning in JavaScript with ml5.js (The Coding Train) and Chapter 10: Neural Networks of The Nature of Code
- p5.js for visualisation and interaction. Recommended: Code! Programming with p5.js (The Coding Train) and The Nature of Code
- Teachable Machine as an accessible route into trainable classification, where appropriate. Recommended: Teachable Machine and the Teachable Machine track (The Coding Train)
For topics such as trust, autonomy and human control, the AI Ethics Framework can help you weigh a system against ethical considerations.