What are the politics of artificial intelligence? What does it mean when we talk about regulating the actions of a machine that expresses intelligence?
"The Politics of AI" report was prepared by GC for a conference in 2019 and covers AI in healthcare, financial services, labour and the economics of data, news and the AI bubble.
In planning our conference themes we have defined AI as a machine system capable of using data analysis to replicate human reasoning, both in achieving certain tasks and in learning from the results of, and adapting, its own protocols. This definition has allowed us to focus on four themes for politics and policy that run through each panel discussion.
The first is the question of what policy frameworks are needed for a world in which machine judgement displaces human judgement. When a machine advises an investor, dispenses a medical diagnosis, drives a vehicle or disseminates a news story, it is undertaking a regulated activity and navigating a human ethical landscape. What ethical challenges do these activities raise? Can we agree on a way to regulate them?
The second is the question of the impact AI will have on the workforce itself. The automation of human work is hardly new: AI is simply the next iteration in the long process of human labour displacement through technology. While unquestionably disruptive, such change is far from inherently negative. But its reach into a wide range of cognition-based jobs – a huge central strata of the modern services economy – will present policy challenges not experienced in the past.
The third relates to the data needed to make the deployment of AI in everyday functions possible. Learning how to think like a person can only be done by analysing the way people think. It is the aggregation for the first time of huge data sets of human behaviour and the computing power to exploit them that has unlocked the current potential in AI. But that data originates in real people, and how it is collected, protected and even remunerated – given its huge potential value – are important questions for policymakers to respond to.
A final theme is the industrial policy question that underlies this change. The race to lead the development of AI is shaping into an international one, but to what extent should it be the subject of industrial policy? To what extent are international frameworks for AI feasible or desirable? And how should policymakers balance support for a lead in the technology itself with adequate consideration of the policy challenges it is raising as it develops?
This conference is intended to debate these questions rather than answer them definitively. It is surely too soon to agree on answers to any of them. But, as this technology begins to transform the world of commerce and work, it is important to ask how it will also impact our politics, and demand sophisticated solutions from both businesses and policymakers.