Leading AI safety researcher Roman Yampolskiy has warned of a potential AI takeover, suggesting there is a 99.9% chance the technology could pose an existential threat to humanity within the next 100 years