I think you're badly underestimating the amount and nature of the data required. Think about space travel: after 60 years, the catastrophic failure rate is stil

I think you're badly underestimating the amount and nature of the data required. Think about space travel: after 60 years, the catastrophic failure rate is still about 1 in 200. This is due in part to the very low volume of flights over the decades. The Shuttle underwent what passed for extensive testing before its first "operational" flight, but was retired before it accumulated even as many hours as the average test flight program for a commercial airliner. Before we run vehicles as potentially destructive as a fully loaded 18-wheeler autonomously on public roads, we should log tens of thousands of test hours under simulated real-world traffic conditions. That's expensive, hence Uber's willingness to roll the dice with other people's lives. The burden is not on the public to show that an autonomous truck is unsafe; the burden is on the promoters to show that it is safe. That burden is not even close to being met, as the article acknowledges when they say the technology won't be ready for another 5 years. This was a stunt to help Uber maintain its valuation, and the DoT should've been looking out more for the public.