REP Research Seminar - Presenting in person - Professor Adam Dennett, University College London. Title: Can we model our way out of this? AI, collective intelligence, and the Brighton & Hove school system.

Abstract: By 2033 Brighton & Hove will have around 760 fewer children entering secondary school each year than it did at its 2019 peak. Schools are funded per pupil, and the decline is not spread evenly. The city has no published long-term plan for it. Instead, there has been a sequence of public consultations about related issues, each proposing a single option and none checking what the last one did.

So I built a working simulation of the whole system: where every child lives, how families actually choose a school, and what each policy lever does to every other school. It reproduces the council’s real admissions outcomes with an R² of 0.95, and it is built entirely from published data — no pupil records, no special access.

The mathematics is fifty years old. What changed is that AI collapsed what should have been months of work into evenings. I’ll demo it live, and show how four modest policies, stacked, take a seriously declining school from 43% full to 94%. None of the policy levers would survive a council committee on its own, so this is evidence that problems relating to complex systems need tools to help model the complex interactions within.

The talk ends somewhere more challenging that I don’t have a solution for yet. If models are now cheap, why do public consultations like those we have experienced over the last 3 years in Brighton and Hove still test consent for existing ideas rather than options? And what would it take for an authority to model honestly, in public, before it has decided?

Bio: I am Professor of Urban Analytics at the Bartlett Centre for Advanced Spatial Analysis, University College London, where I have been based since 2010, first as a post-doc and then as a Lecturer and Senior Lecturer / Associate Professor. I became Professor in 2021. From 2018 to 2023 I was Head of Department and I am an Elected Fellow of the Academy of Social Sciences as well as a Fellow of the Royal Geographical Society.

My academic interests are broad and in the past I have described myself as a bit of a ‘magpie’; however, I have published on topics such as migration and residential mobility (and particularly modelling migration using spatial interaction models), housing, gentrification, geographies of beer and brewing, synthetic data and longitudinal analysis. More recently I’ve been delving into schools data.

Tying all of these themes together is a focus on data and quantitative methods — especially if I can use R — and a passion for applied work which often finds a home in urban settings. I’m a geographer, so I’m contractually obliged to always look at the spatial dimension of any problem and I’m yet to find a situation where it doesn’t matter in some way.