Assessing the Singularity

Sammanfattning

Artificial Intelligence (AI) has had profound growth over the last decade. Will there be an AI system in the future that reaches a stage where it outputs a better AI system? David Chalmers Singularity argument suggests so. The first premise, the equivalence premise, states that absent of defeaters, we will build an AI system at human level intelligence. The extension premise says that we can extend a human level AI to superhuman level. The amplification premise says that the superhuman AI can create a more intelligent AI until it reaches orders of magnitude beyond humans. In this essay, I will defend the singularity thesis while also arguing that the speed of the singularity might be slow, on the order of a decade. I also argue that a singularity is potentially of human extinction risk from a such an amplification.