Teaching, Lectures and Conferences

My academic work has been fortunate to gain recognition, with citations in books and refereed journal articles, and inclusion in teaching materials and reading lists for both undergraduate and postgraduate courses around the world. I contribute to the British Psychological Society course for its Certificate of Occupational Test Use and have the privilege of being involved in the MSt in AI Ethics and Society and the MPhil in Ethics of AI, Data, and Algorithms at the Centre for the Future of Intelligence, University of Cambridge.

That said, not all of my work is as easily accessible—such as keynote presentations, public lectures, and conference talks. While these are harder to track down, I’m pleased to share further details and materials, such as PowerPoint slides and full texts, below.

Smart Teaching, Smarter Assessments: The GenAI Revolution

This is the era of Generative AI. As it gets underway, all will change, so we should soon be able to progress to more interactive educational styles, even though it is difficult now to foresee how these will look. Preliminary exploration of the potential of GAI in the creation of innovative teaching in psychometrics proved that the possibilities were beyond our wildest dreams! Not only could GenAI create the test specification, suggest items, respond to prompts to make changes in items (e.g. Rust, J. (2025), Pellert et. al. 2024) (Vahid Aryadoust et al) , it could also critique the design and draft narrative reports (an otherwise cumbersome task). These are all

Curriculum-Anchored Provisional Norms

During the early development of a new ability test, full age-standardised norms will not yet be available. This does not mean, however, that nothing is known about the level of performance that can reasonably be expected from children of different ages. National curricula are themselves constructed around assumptions about what an average child should have learned, understood and become capable of doing at particular ages or school stages. Teachers also possess substantial practical knowledge of which questions children at different levels are ordinarily able to answer.

Curriculum expectations, teacher judgement and evidence from previously standardised ability and attainment tests can therefore be used to construct provisional estimates of expected item difficulty, average performance and likely variation within an age group. Such estimates would not constitute final psychometric norms, nor would they remove the need for representative standardisation. They could nevertheless provide a useful interim framework for research, automated item generation, initial item selection and the construction of preliminary test forms.

Because contemporary AI systems can analyse national curriculum documents, attainment targets and examples of age-appropriate work, they can assist in drafting large pools of candidate test items at provisionally specified difficulty levels. The curriculum can supply the substantive knowledge and skills expected at each age or school stage, while psychometric principles can guide item format, distractor construction and progression in difficulty. These AI-generated items would still require expert review and empirical piloting, but curriculum anchoring could make item generation considerably faster, more systematic and better targeted than beginning without any prior estimate of age-appropriate difficulty.

The idea arose from my work on the standardisation of the Wechsler Intelligence Scale for Children. Test development need not begin from a position of complete ignorance about age-related ability. Existing educational knowledge can be used to generate and order items provisionally, after which pilot data, item-response analysis and ultimately a full representative standardisation can progressively replace judgement with empirical estimates..

Individual Tuition

But as well as the process of test construction itself there are many other opportunities. As teaching methods themselves begin to develop in line with the new possibilities, we expect to see major advances in the development of individual tuition. GenAI is, after all, not an individual – it doesn’t have a ‘self’ in the way humans’ do. But because of this, a single GenAI is capable of carrying out a very large number of individually tailored instructions simultaneously to a large number of learners. And as each learning is completed it can be analysed, responses assessed and new instructions prepared – all on an individual basis.

What the future holds

This all paints a completely different picture of the future. For assessment of learning, not just in the school classroom, but also online, at home and in university. It also clearly demonstrates the need for more psychometric training for the new brand of professionals that will be required. In terms of its ability to ‘understand’ as well as predict human psychology and behaviour, it also opens up possibilities for augmenting our own understanding of human psychology. Given the current state of our interconnected world, this is now more important than ever.