Module manager: Jennifer Sexton
Email: J.L.Sexton@leeds.ac.uk
Taught: Semester 1 (Sep to Jan) View Timetable
Year running 2026/27
Students with a BSc in Computer Science at 2.1 (or equivalent) can take either this module or Advanced Computer Science.
This module is not approved as an Elective
This module explores the history and development of AI, showing how ideas from applied mathematics and statistics were adapted into the core concepts behind contemporary AI systems. Students will learn about the range of statistical and computational approaches commonly labelled as AI, from pattern recognition and prediction to text generation and automation, and how they relate to each other. The module takes the form of interactive workshops exploring real-world case studies. Through dashboards, applications, and analytical tools, students will investigate technical limitations of AI systems including: overfitting and generalization failures, distribution shift across deployment contexts, class imbalance and fairness trade-offs, algorithmic bias through proxy variables, sample size constraints for rare events, measurement error and data quality issues, and the gap between AI safety claims and empirical reality. Each concept is grounded in concrete examples of AI system failures and their governance implications. The module requires no prior programming experience or mathematical background. Teaching is structured around active learning: before each workshop, students will explore interactive tools independently; during workshops, students will analyse real cases collaboratively; after workshops, students will apply concepts through guided exercises. This scaffolded approach builds technical literacy progressively, moving from observation to analysis to independent evaluation. Formative feedback throughout helps students develop confidence in engaging with technical material and translating it for policy contexts. This technical foundation enables critical evaluation of AI systems alongside the ethical, legal, and political analysis developed in subsequent modules. By the end of the module, students will be equipped to interrogate AI vendor claims, recognize systematic failure modes, and contribute to evidence-based AI governance and policy design.
This module demystifies AI by grounding it in established mathematical frameworks, revealing both the capabilities and fundamental limitations arising from data quality, algorithmic constraints, and statistical assumptions.
Students will use interactive dashboards, applications and tools to explore real-world case studies that demonstrate how these limitations manifest in deployed AI systems. Students will develop the critical literacy skills needed to evaluate claims about AI systems and identify what is technically possible.
By the end of the module, students will be equipped to analyse AI vendor claims and recognize the causes of common systematic failures and whether the risks of these failures can be quantified or mitigated via evidence-based AI policies and governance.
On successful completion of the module students will be able to:
1. Describe the statistical concepts underlying AI systems, including sampling, probability distributions, correlation vs causation, and uncertainty quantification;
2. Explain fundamental limitations of contemporary AI systems, including computational constraints, generalization failures, and the gap between training and deployment conditions;
3. Identify systematic failure modes in AI systems (e.g. overfitting, class imbalance and provenance destruction) and their implications for AI governance;
4. Communicate the technical limitations of AI systems to non-specialist audiences by translating statistical concepts into clear governance-relevant analysis. (Work ready, Digital, Sustainability);
5. Critically evaluate AI system performance claims and assess whether empirical evidence supports vendor assertions. (Work ready, Sustainability, Academic);
6. Integrate technical analysis with ethical, legal, and political frameworks by proposing policy interventions grounded in realistic understanding of AI capabilities. (Sustainability, Enterprise).
| Delivery type | Number | Length hours | Student hours |
|---|---|---|---|
| Seminar | 2 | 1 | 2 |
| Seminar | 9 | 2 | 18 |
| Private study hours | 130 | ||
| Total Contact hours | 20 | ||
| Total hours (100hr per 10 credits) | 150 | ||
130
Resitting students will be allowed to rework their previous submission and include a summary of the changes they have made. The exception to this is students resitting due to academic integrity issues who will be required to select a new sector and/or application for the case study.
| Assessment type | Notes | % of formal assessment |
|---|---|---|
| Coursework | Coursework | 100 |
| Total percentage (Assessment Coursework) | 100 | |
Normally resits will be assessed by the same methodology as the first attempt, unless otherwise stated
Check the module area in Minerva for your reading list
Last updated: 30/04/2026
Errors, omissions, failed links etc should be notified to the Catalogue Team