The report is also concerned with the possibility of exponential advances in artificial intelligence. Once computer programs grow advanced enough to teach themselves computer science, they could use that knowledge to improve themselves, causing a spiral of ever-increasing superintelligence.
•If AI remains friendly to humans, this would be a very good thing indeed, and has the prospect to speed up research in a variety of domains. The risk is that AI has little use for humans and either out of malevolence or perceived necessity destroys us all.