Human Technology Institute researchers are developing new AI methods to help policy makers and communities address complex social issues.
HTI researchers are working with government departments, philanthropic institutions, community groups and other research organisations to try to untangle the interconnected pathways that lead to longstanding social challenges and identify what matters most to communities and which initiatives work best to alleviate social disadvantage.
Thanks to huge increases in computational power and the development of new mathematical methods and algorithms, we are able to solve mathematical formulae that have existed for hundreds of years to find answers to questions that until now have been intractable.
The new algorithms that HTI researchers have developed are capable of computing the probability that a particular causal pathway, out of billions and billions of possible pathways, leads to outcomes which matter to communities. This is enormously important because if we understand what leads to outcomes it is possible to ask what initiatives work that enable communities and individuals to overcome barriers.
These new methods synthesize and learn from a range of different data sources, including the views of impacted communities; they adapt as new information comes to hand; they shed insights on which initiatives work best; and give rapid feedback on the effectiveness of these initiatives – promoting a learning mindset while maintaining technical rigour. This will allow communities and governments to draw conclusions and make collective decisions on issues with multiple and complex interacting factors and causes.
HTI researchers are developing these methods for a range of complex social issues including to understand the drivers behind and solutions to youth mental health; and breaking cycles of intergenerational disadvantage for parents of young children. These methods could be used for countless other issues facing governments, such as how do we manage our natural resources and which initiatives may alleviate the housing crisis.
The Thrive cycle for adaptive learning
In this series we’ll hear from a number of people from HTI and outside HTI speaking at the Data for Policy Conference 2024 who will talk about why we need this approach, how it works, and how it could be applied to other areas of government policy.
Using an example from the Thrive: Finishing School Well program, HTI’s Professor Sally Cripps explains in detail the methodology for adaptive learning.
Which policies do we implement to tackle social disadvantage? In an excerpt of a presentation prepared for the Data for Policy Conference 2024, Professor Cripps explains why we need policies and programs that are adaptive, combine multiple information sources and reflect community values.
Sir Geoff Mulgan from University College London discusses the current challenges of government in solving complex problems when our institutions and approaches are siloed, rather than collaborative.
Dr Stefaan Verhulst from GovLab explains why we need diverse decision makers – and the current challenge of not having the right decision makers in the room or understanding when we have overlapping decision responsibilities.
Ben Gales from the Paul Ramsay Foundation discusses the limits of traditional evaluation methods such as randomized control trials (RCTs), which are done after a program is finished – when it’s too late to make changes to see if other approaches would have worked better.
To overcome these shortcomings with traditional research methods, Professor Sally Cripps explains how Bayesian Adaptive Learning works, using an example from a study on the causes of childhood obesity.
Copyright © 2024 Thrive: Finishing School Well