Foundations of Human-Centred AI - 2025 entry

 

You will study foundational concepts in how to design Artificial Intelligence (AI) systems that interact with humans. This will involve learning about human psychology including computational theories of how people represent and process knowledge, learn and work together. You will learn about topics including, how people make decisions, how they perform perceptual/manual tasks, how human vision works. You will use these theories to build and critically evaluate Artificial Intelligence systems that work with people.

 

You should take this module if you are interested in going on to a masters/research degree and/or in the rapidly expanding number of career pathways that involve designing AI to work with people. For these careers learning about human psychology is vital to designing systems that, for example, people find useful but not controlling and people find engaging but not addictive. For example, answers to the following questions require an understanding of the psychology of the user. How can AI be fine-tuned to human preferences and emotions? How can an AI system learn about an individual person’s goals and preferences? How can it learn about their emotions and feelings about others? Answers to these questions can help improve AI systems that work with people in the workplace and the home.

 

You will attend a weekly class in which an expert in Human-centred AI will lead discussions about a particular topic. You will work individually and in groups to investigate assigned topics and present your work.

 

Some mathematics and Python knowledge is needed for this module. No prior knowledge of human psychology is required

 

The module is recommended for interdisciplinary pathways.  

 

The module will cover topics such as recommender systems, emotion detection systems and decision support systems.  It will also cover topics including how to model humans with computer programs using techniques such as artificial neural networks, reinforcement learning with human feedback (RLHF), Bayesian inference, Large-Language Models, generative AI, optimisation and game theory. These are topics that are vital to Human-centred Artificial Intelligence. For example, RLHF is used to fine-tune Large-language models, through interaction with humans, so that the models provide information that people find useful.

 

The module will be informed by the latest research in the rapidly developing field of Artificial Intelligence and Human-Computer Interaction (HCAI). It will support careers in organisations that are seeking to incorporate AI into various aspects of work including automation of routine tasks, decision support and insights, personalisation of work tools, collaboration and communication, creative and strategic applications. It is relevant in a wide range of industries including finance, manufacturing, education and healthcare.

The module will cover the following indicative topics:

 

• Recommender systems

• Emotion detection systems

• Decision support systems

• Computational modelling of human behaviour

 

We will explore these topics in context of state-of-the-art AI techniques, including deep reinforcement learning, Bayesian inference, optimisation, and game theory.