An AI approach to (public health) policy.
When a pandemic strikes, policy-makers face a complex decision about what intervention(s) are best for society — both in terms of lives and dollars. The calculations are too complex and emotionally-fraught for human policy-makers to perform well.
I propose a future where an AI policy-maker analyzes these scenarios, then suggests (and possibly enacts!) changes to public policy in real-time, responding dynamically based on what results the policy has. We outline a technical approach here.
This philosophy applies beyond pandemics, to other policy issues (e.g. climate change). Such scenarios require policy change both as the problem changes (virus mutates, CO2 levels rise), and as new tech becomes available (vaccines developed, alternative energy sources come online). Human decision-makers don’t respond quickly enough and can’t process enough channels of information. Our political process makes things even less rational.
I envision a world where AI negotiators help side-step the failings of politics and public discourse, by presenting well-reasoned motivations for their decisions, and by offering appropriate compensation to those who would be hurt by a policy that otherwise maximizes social good.
One can view this as an extension of Posner and Weyl’s Radical Markets proposals. We need AI to propose and implement these markets, identify the opportunity, and build the software to run them.
[Addendum: Note that this idea is starting to enter the mainstream, with the likes of the Boston Consulting Group advocating for some form of AI-based policy-making.]