CSC8802 : Safe and Trustworthy AI
- Offered for Year: 2026/27
- Module Leader(s): Dr Varun Ojha
- Lecturer: Professor Shishir Nagaraja, Dr Tejal Shah
- Owning School: Computing
- Teaching Location: Newcastle City Campus
Semesters
Your programme is made up of credits, the total differs on programme to programme.
| Semester 2 Credit Value: | 10 |
| ECTS Credits: | 5.0 |
| European Credit Transfer System | |
Aims
With the near ubiquitousness of AI technologies and that these operate in complex, real-world scenarios, there is growing potential for AI to do harm and erode trust with serious consequences in critical systems. Building and deploying safe AI systems is therefore vital to minimise risks and ensure behaviour of AI systems is aligned to societal ethics and values. This module will introduce the important topic of Safe AI covering practical tools and technologies for building, deploying, and testing safety of AI systems. It will also cover AI safety from ethical, regulatory, and social perspectives – exploring frameworks for building safe AI systems, including bias mitigation strategies, regulatory safeguards, and societal impact of AI technologies. Students will be able to work in AI Safety lab jointly created with Lenovo, thereby gaining industry collaboration, real-world project options, and subsequent internship opportunities (offered through joint Edge AI Hub, Lenovo, SPARK, and other industry partnerships). Further equipment (robots) will be supplied by SPARK so students can gain experience of testing for safe operations in different operational contexts.
Outline Of Syllabus
Following is a broad overview of the topics that will be covered in this module:
• Fundamental concepts of AI safety, trustworthiness, fairness, robustness, and accountability.
• Tools and techniques for building safe and trustworthy AI systems.
• Security of AI.
• Problems in AI Safety: Robustness and distributional shift; Scalable oversight and Safe exploration; Negative
side-effect and Reward hacking.
• Safety assurance and evaluation methods.
• Explainable and interpretable AI.
• Generative AI and hallucinations.
• Foundational models and transfer learning practices.
• Human oversight and control.
• Ethical and regulatory frameworks for safe AI.
Teaching Methods
Teaching Activities
| Category | Activity | Number | Length | Student Hours | Comment |
|---|---|---|---|---|---|
| Scheduled Learning And Teaching Activities | Lecture | 4 | 1:00 | 4:00 | Synchronous, online sessions. |
| Guided Independent Study | Assessment preparation and completion | 1 | 30:00 | 30:00 | Assessment preparation. |
| Scheduled Learning And Teaching Activities | Lecture | 4 | 2:00 | 8:00 | Lectures (in person) on core concepts. |
| Scheduled Learning And Teaching Activities | Practical | 4 | 2:00 | 8:00 | Practical (in person), hands-on activities. |
| Guided Independent Study | Directed research and reading | 5 | 2:00 | 10:00 | A flipped classroom approach, students will undertake research/activity individually and asynchronously. |
| Scheduled Learning And Teaching Activities | Drop-in/surgery | 4 | 1:00 | 4:00 | Weekly optional drop-in surgeries. |
| Guided Independent Study | Independent study | 1 | 36:00 | 36:00 | Independent study on course content. |
| Total | 100:00 |
Teaching Rationale And Relationship
Teaching methods include a combination of lectures covering the fundamental concepts and theoretical
knowledge of AI Safety and Trustworthiness. Practical sessions will allow students to apply this knowledge to
designing and working with solutions that make AI safe. A flipped classroom approach will encourage students to
come prepared for the in-person/synchronous sessions, which would include interactive discussions and active
participation to reinforce student learning.
Assessment Methods
The format of resits will be determined by the Board of Examiners
Other Assessment
| Description | Semester | When Set | Percentage | Comment |
|---|---|---|---|---|
| Report | 2 | A | 100 | A report covering theoretical concepts and practical skills covered in the module. |
Formative Assessments
Formative Assessment is an assessment which develops your skills in being assessed, allows for you to receive feedback, and prepares you for being assessed. However, it does not count to your final mark.
| Description | Semester | When Set | Comment |
|---|---|---|---|
| Report | 2 | M | A draft smaller report for feedback. |
Assessment Rationale And Relationship
The report will assess students’ ability to apply principles to assess and design safe and trustworthy AI-driven solutions. The smaller report will provide feedback to students.
Reading Lists
Timetable
- Timetable Website: www.ncl.ac.uk/timetable/
- CSC8802's Timetable