Because the immediate downside risk is small and the upside benefit is incredible. Even with very low confidence in self driving vehicle development, eg. 1% chance of success in 10 years, the upside is overwhelmingly net positive in terms of lives saved, on top of all the other benefits of automating transport like freeing almost 10% of human drivers' waking hours in the US. If the risk taken was anywhere close to what it would have to be to scale down at the cost of slower deployment for safety reasons, we would know.