Are self-driving cars safer than human drivers?

A discussion about AI reliability turns into a dispute over driving safety. One participant calls current AI agents “repeat offenders” at failing to follow instructions, contrasting them with reliable software such as spreadsheets. A reply claims self-driving cars are much safer than any form of human driving and extends that claim to other AI systems. The critic rejects the comparisons as unfair, arguing that safety in unfamiliar environments is a major reason driverless cars remain in limited deployment. The proponent counters that empirical evidence overwhelmingly shows AI-driven cars reduce fatalities.