AI NEEDS YOU

Regulating AI will not be easy, but there are useful lessons available.

Recently, a slew of books have warned about the dangers of AI, so it is refreshing to find one that takes a different approach. Harding has been working at the intersection of technology and regulation for many years; she used to direct policy for Google DeepMind, and she is now director of the AI & Geopolitics Project at the Bennett Institute for Public Policy at the University of Cambridge. The author has a positive attitude toward AI, and she walks us through the benefits already gained or on the horizon. She also acknowledges the downsides. AI has reached the stage of requiring a global regulatory framework, she notes, perhaps begun by technical experts but eventually incorporating input from other interested parties. There are instructive precedents for this, such as the UN Outer Space Treaty, which has been effective at preventing the deployment of orbiting weapons. The regulation around in vitro fertilization is another example of workable compromises involving transformative technologies. A third area to examine might be the protocols drawn up to organize the internet, which have been remarkably durable. All of these required time and effort, with conferences and written agreements. Harding believes that the U.S., as the leader in the field, has to kick-start the process. She is likely correct, but it is hard to see how global talks could take place in the current heated geopolitical climate. Russia is not willing to discuss anything, and China has clear ambitions to dominate the AI area. Nevertheless, we might hope that this book ends up in the hands of those who could take a leadership role, before it is too late.

Harding’s optimism is tempered by experience and knowledge, and her arguments are compelling and engaging.