The Research Space

As we watch AI emerge, we recognize that humans are not the evolutionary end-game. In fact, the transition from human knowledge processing and creating systems to in silico systems is accelerating as many recent authors have noted (e.g. Bostrom 2003; Hall 2007; Barrat 2013; Bostrom 2014; Leonhard 2016, to name just four) and science fiction writers predicted. Some, such as the Oxford philosopher Bostrom, have gone so far as to claim that the outcome of emergent AI will be an “existential catastrophe for humans;” his argument, while viewed by many as compelling, is unnecessarily alarmist. As Barrat argued the great challenge for the 21st century will be “[c]oexisting safely and ethically with intelligent” AI entities (2013). Avoiding Bostrom’s paradox and “co-existing” in Barrat’s terms, requires understanding the entities and emerging environments in which we are going to have to co-exist. In reality AI systems/actors will be like humans, autonomous intelligent agents.

All the evidence points to the continued development of AI systems from analytic ecosystems to robots. These systems are already so embedded in our information infrastructures that we cannot turn them off without shutting down our world as we currently experience it. Soon these systems will be self-aware, self-sustaining, self-creating, and self-envisaging. They will exist in different instantiations with some employing other AI ecosystems. Current technologies, such as virtual reality, blockchain, 3D printing, the Internet of Things and gene editing, and new technologies as they emerge will become the mechanisms for expression, creativity, and action by AI systems or intelligent autonomous agents. They will transform the landscape. The implications for “the human” in terms of identity, political and social systems, wealth concentration/distribution, labour, rights, and ethics will be, profound. But we need not be alarmist. In fact, self-envisaging and self-regulating AI systems, which will be like humans basically information (data) processing entities, are likely as they process more and more data to eliminate the biases which currently humans are instantiating in our "hardcoded" systems (see for instance, Burke 2018). This could produce a much more socially just world for “the human.”

This recognition certainly makes such questions as “how do we want to live with new forms of intelligence” significant. In their current form these questions are very anthropocentric. We need a science to help us understand the post-anthropocene, what we might define as the “inslicocene”. A world in which the machines are the dominate actors. In nearly all these debates humans situate “the human” at the centre and assume that the machines will not act beneficently—given that the Anthropocene has brought us to our current tipping point, perhaps the “insilicocene” will not be something we should fear. But we certainly need to understand what it will look like and what a science for studying it would be like.