What are the limits of Artificial Intelligence?

Seldom a day goes by without a new report in the media about artificial intelligence, reporting new achievements and inspiring hopes and fears for the future. But how realistic are these visions?

Barry Smith, well-known contemporary philosopher, with Jobst Landgrebe, artificial intelligence entrepreneur, recently published a book entitled ‘Why machines will never rule the world: Artificial Intelligence without fear.‘ In this book, Landgrebe and Smith argue based on the mathematical theory of complex systems that artificial general intelligence – at the level of human intelligence – will never be possible.

If Landgrebe and Smith are correct, then AI systems will always be limited to offering a simulation of intelligent behaviour rather than genuine intelligent behaviour, and therefore will only be applicable to limited situations which are predictable, rather than the wide range of unpredictable and novel situations that humans navigate in daily life.

We are editing a special issue of the journal Cosmos+Taxis dedicated to responses to the book. The special issue will explore topics such as what the limits of artificial intelligence might be, and what the implications are of these limits for how we design, interact with and regulate the development of such systems. If you are interested in contributing to this special issue, contact me.