What does the automation of imagination via AI-based “imagination machines” mean for the development of imagination as a tool for solving public policy problems?

Fields of human creativity from literature to architecture have seen a wide variety of experiments in the ability of artificial intelligence – self-learning algorithms fed with big data sets – to imagine. Sridhar Mahadevan speaks of these algorithms as ‘imagination machines’.

Good public policy, at any level, requires that we are able to imagine new ideas and yet, often, short-term budgets, professional pressures, and political cycles make it difficult to think outside of the immediate present. Can AI-based imagination machines help us overcome those obstacles? It is often said that AI does not decide for us, but instead provides only a series of evaluations and options from which humans ultimately get to serve as the decision-makers. However, if humans aren’t actively involved in the imagination of alternative policy suggestions, do they remain capable of understanding how to make appropriate decisions? How can we ensure public accountability if decision-makers don’t understand the decisions they make? And, how can we at the same time ensure that we can achieve the potential AI has to help us find new ways to imagine a better society?

With Professor Sridhar Mahadevan, University of Massachusetts, Director at Adobe Research in San Jose, California, USA.

Recorded on 12 April 2021. This seminar is part of a series hosted by the research platform Collaborative Future-Making.