If artificial intelligence (AI) were achievable, what would the consequences be for human society?1 Perhaps surprisingly, the answer to this question is already at hand. We are achieving rapid and accelerating success in our quest to build AI. That very success — and the slowness with which both the academic community and the general public have come to recognise it — has shown how little we understand our own intelligence, and its role in our lives and culture. Here I attempt to address this problem of understanding, exploiting a variety of scientific evidence, including social simulation. I begin by reviewing current progress in AI, which is profound but underestimated. I suggest this lack of recognition is due to the mistaken belief that intelligence implies agency. I next examine the related question of human uniqueness: why do only we have language and extensive built culture? I use models and data to show that the propensities to use culture, share information and behave altruistically are neither unique to humans nor inexplicable to biology, but rather our uniqueness hinges on the extent of our capacities for communication and memory. Finally, I apply the impact of AI on extending our intelligence to these theories, to predict—and observe—consequences of AI on human societies and individual human lives. I make and support policy recommendations based on these predictions.
|Title of host publication
|Collective Agency and Cooperation in Natural and Artificial Systems
|Subtitle of host publication
|Explanation, Implementation and Simulation
|Place of Publication
|Published - Oct 2015
- artificial intelligence; levels of selection; agency; human behavioural ecology; cognition