There are a number of issues that need to be considered when it comes to self-driving cars, including consumer trust, crash data, urban planning implications, and the environmental costs of self-driving technology. These issues are explored in relation the U.S.
Autonomous vehicles (AVs), or self-driving cars, are vehicles equipped with advanced sensors, cameras, radar, and AI that allow them to navigate and operate without human intervention.
The findings are drawn from CR Legal Team’s autonomous vehicle safety study, which assesses the long-term road safety implications of full autonomous deployment.
What do drivers really think?
Interest in self-driving cars is greater with the younger generation. For example, 51% of Gen Z say they would ride in a self-driving car, which is the highest of any generation.
However, overall trust is low. Only 13% of all US drivers trust self-driving vehicles. The generation most primed to adopt this technology is, in most cases, still saying no.
Crash rates remain high
Autonomous vehicle crashes reported to NHTSA nearly doubled in 2024, from 288 to 544, but the fleet expanded significantly over the same period, and just one fatality has been recorded across all fully autonomous vehicle crashes to date.
The number measures deployment growth more than it measures risk. Where the technology has been deployed, crash rates have already fallen from 9.1 per million miles to 4.6.
Yet safer driving is forecast
By 2050, projections suggest self-driving cars could save around 21,700 lives and prevent 4.22 million accidents and save 21,700 lives every year, because 94% of all US crashes involve human error, and autonomous systems do not get tired, drunk, or distracted.
Autonomous vehicle testing has already decreased the national crash rate from 9.1 per million miles to 4.6 per million miles, a near 50% reduction in accidents where such technology has been deployed.
Urban impact
Self-driving cars could reduce urban parking demand significantly, but only if vehicles are shared, pooled, metro-wide, and paired with public transit. Remove any one of those conditions and the benefit shrinks to near zero, and full fleet automation is not projected before 2045.
Estimates suggest that one billion autonomous vehicles running one hour a day on current onboard computers would generate greenhouse gas emissions equivalent to every data centre in the world today. The vehicles may produce less pollution at the tailpipe. The computing infrastructure behind them could more than cancel that out.
Overall assessment
The report indicates that the points above are not fringe concerns or distant projections. They are the central tensions in a technology that is already on public roads and already shaping policy, investment, and planning decisions across the country.
