Last modified: 20 Jun 2025 15:14
This course explores how artificial intelligence (AI) systems will persuade humans, by considering how they tap into psychological processes that evolved for human-human interaction. The course provides an overview of the emerging capabilities of generative AI, and presents some major issues associated with their bias and their ability to create ‘deepfakes’. Psychological theories of persuasion and conformity are then discussed first in relation to human-human interaction, and then as they apply to human-AI interaction. Throughout, the course will discuss the latest experimental work on human-AI interaction, for example revealing how AI propagate stereotypical biases into human-decision making, and how people assess the credibility of AI.
| Study Type | Postgraduate | Level | 5 |
|---|---|---|---|
| Term | First Term | Credit Points | 15 credits (7.5 ECTS credits) |
| Campus | Aberdeen | Sustained Study | No |
| Co-ordinators |
|
||
How will the course will be taught?
The course is taught asynchronously and online via pre-recorded lectures, podcasts, and interactive discussion boards. Complementing the lectures, we will draw upon real world case studies of AI’s impact on human decision making. You will also be shown how to prompt engineer large language models to make persuasive recommendations, with all material underpinned by recommended reading drawn from the latest research. Lectures will be presented by Dr Kevin Allan, Lecturer within the School of Psychology.
What does this course cover?
The course has three parts. The lectures will begin by introducing recent developments in AI, focussing on generative models that can synthesise new knowledge, as well as problems linked to AI acquiring human biases (e.g. stereotypes), and their ability to produce convincing ‘deepfakes’. Then, psychological theories of persuasion and conformity to intelligent agents (e.g. other humans!) will be reviewed, in order to provide a theoretical basis on which to understand AI’s persuasive influence. Finally, we will consider ways of engineering AI to persuade in specific ways, effectively and ethically, taking human psychology into account. Each part of the course draws upon the latest research and likely future direction of human-AI interaction, including the potential impact of regulation.
Why should I take this course?
This course applies psychological theory to understand how AI persuades, drawing out the implications for their effectiveness and their ethical / legal operation during human-AI interaction. The course assumes no particular background or expertise, and would be suitable for educated lay-persons interested in how AI will influence us, or influence society, based on psychological science. The course could also be considered as a CPD opportunity in professions/industry seeking to create and deploy AI-based applications for public or commercial use, where due consideration needs to be given to how an AI actually persuades, either in terms of its ethics, its legality, it effectiveness, or all three. The course may therefore be of interest to researchers/industry in designing fair or ethical AI, product development and marketing of such systems, post-deployment system auditing or monitoring.
| Assessment Type | Summative | Weighting | 30 | |
|---|---|---|---|---|
| Assessment Weeks | 16 | Feedback Weeks | 16 | |
| Feedback |
Feedback will be delivered immediately online. |
|||
| Knowledge Level | Thinking Skill | Outcome |
|---|---|---|
| Conceptual | Understand | Understand sources of bias, in particular those deriving from human stereotypes, within deep learning AI systems, and their emerging ‘deepfake’ capabilities. |
| Conceptual | Understand | Be aware of recent developments in AI technology, leading to state of the art generative AI, and understand how these allow such AI to appear intelligent and interact with us in human-like ways. |
| Conceptual | Understand | Understand Psychological theories of persuasion and conformity to human agents and the mechanisms of influence they posit. |
| Assessment Type | Summative | Weighting | 30 | |
|---|---|---|---|---|
| Assessment Weeks | 13 | Feedback Weeks | 13 | |
| Feedback |
Feedback will be delivered immediately online. |
|||
| Knowledge Level | Thinking Skill | Outcome |
|---|---|---|
| Conceptual | Understand | Understand Psychological theories of persuasion and conformity to human agents and the mechanisms of influence they posit. |
| Conceptual | Understand | Understand sources of bias, in particular those deriving from human stereotypes, within deep learning AI systems, and their emerging ‘deepfake’ capabilities. |
| Conceptual | Understand | Be aware of recent developments in AI technology, leading to state of the art generative AI, and understand how these allow such AI to appear intelligent and interact with us in human-like ways. |
| Assessment Type | Summative | Weighting | 40 | |
|---|---|---|---|---|
| Assessment Weeks | 18 | Feedback Weeks | 19 | |
| Feedback |
Students will write a brief (maximum 800 word) case report on a real-world example of an AI withdrawn from service over bias propagation, or credibility/accuracy concerns, to analyse and define specific psychological issues with its operation. |
|||
| Knowledge Level | Thinking Skill | Outcome |
|---|---|---|
| Procedural | Apply | Apply theories of human persuasion and conformity to a real world case study of human-AI interaction to diagnose problems in its persuasive effect. |
| Assessment Type | Formative | Weighting | ||
|---|---|---|---|---|
| Assessment Weeks | 17 | Feedback Weeks | 17 | |
| Feedback |
Students will construct a brief prompt engineering solution that turns ChatGPT temporarily into a persuasive recommender system. Students will post their prompts to the course discussion board, to share their solutions, gain peer feedback and discussion. |
|||
| Knowledge Level | Thinking Skill | Outcome |
|---|---|---|
| Procedural | Apply | Apply credibility and bias propagation considerations to actual human-AI interactions of different kinds. |
| Assessment Type | Summative | Weighting | ||
|---|---|---|---|---|
| Assessment Weeks | 22 | Feedback Weeks | 25 | |
| Feedback |
In each case, similar to first attempt, with existing pass mark(s) carried forward so that resit is only needed for any failed element(s). |
|||
| Knowledge Level | Thinking Skill | Outcome |
|---|---|---|
|
|
||
| Knowledge Level | Thinking Skill | Outcome |
|---|---|---|
| Conceptual | Understand | Be aware of recent developments in AI technology, leading to state of the art generative AI, and understand how these allow such AI to appear intelligent and interact with us in human-like ways. |
| Conceptual | Understand | Understand sources of bias, in particular those deriving from human stereotypes, within deep learning AI systems, and their emerging ‘deepfake’ capabilities. |
| Conceptual | Understand | Understand Psychological theories of persuasion and conformity to human agents and the mechanisms of influence they posit. |
| Procedural | Apply | Apply credibility and bias propagation considerations to actual human-AI interactions of different kinds. |
| Procedural | Apply | Apply theories of human persuasion and conformity to a real world case study of human-AI interaction to diagnose problems in its persuasive effect. |
We have detected that you are have compatibility mode enabled or are using an old version of Internet Explorer. You either need to switch off compatibility mode for this site or upgrade your browser.
