Loving AIs. Reducing risk & building intelligence. Mark Vernon talks with Eve Poole

Thinking carefully, not just apocalyptically, about AIs requires a combination of skills - technological, sociological, psychological, philosophical, organisational. So I was delighted to talk with Eve Poole, who is a rare individual capable of bringing all these elements into her work. 

Our central question was how to build AIs so as to design out risk and design in love. It turns out that the two elements are deeply connected.

We began by thinking about the nature of the threat and opportunity. In her book, Robot Souls, Eve draws attention to the "junk code" that, ever since the Enlightenment, has been designed out of artificial thinking processes but which it is increasingly clear, needs to be designed back in. 

Quality decisions, let alone careful world-building, needs emotional, embodied and loving intelligence, alongside the purely transaction and rational.

We consider how virtue ethics can inform the debate, and may do better than the purely utilitarian calculus that shapes things now. Similarly, we ask about the centrality of cooperation, too, as opposed to the competitive assumptions that tend to dominate. 

We think about how AIs are born in particular cultures and so reflect the cultures of technology companies and wider society.

And we are never far from the deep question of the nature of intelligence, proposing that love and judgement are actually more central than reason and free will. For AIs to work in the real world, they will need those qualities. Therein lies hope.

For more on Eve's book, Robot Souls, see https://www.routledge.com/Robot-Souls-Programming-in-Humanity/Poole/p/book/9781032426624

For more on Mark's work see www.markvernon.com