A new study says that waiting too long to roll out self-driving cars will lead to thousands of preventable highway fatalities.

The study by the RAND Corporation is sure to stir an already heated debate among automakers, regulators, and safety advocates about how safe self-driving cars should be before they’re widely allowed to roam public roads. RAND says that introducing the technology when it’s only a little better than human drivers will save the most lives.

“If we wait until these vehicles are nearly perfect, our research suggests the cost will be many thousands of needless vehicle crash deaths caused by human mistakes,” says Nidhi Kalra, co-author of the study and director of RAND’s San Francisco office. “It’s the very definition of perfect being the enemy of good.”

The study comes as both Congress and the National Highway Traffic Safety Administration are reviewing their policies in an effort to ease the introduction of self-driving cars. Bills that would enable automakers to expand testing of automated vehicles on public roads are nearing approval. The proposals would also permit public sale of hundreds of thousands of advanced-technology cars that would be exempt from existing federal safety standards.

More on Self-driving cars

In creating the study, RAND evaluated the likely outcomes of three scenarios: quick adoption of self-driving cars when they’re initially just 10 percent better than human drivers, delaying rollout until they are 75 percent better, and delaying until they’re 90 percent better. The faster introduction of the less perfect systems would save thousands of lives over the next 15 years and possibly hundreds of thousands over the next three decades, the study found.

The RAND analysis echoes arguments made by some auto executives, such as Tesla CEO Elon Musk, who has said it’s morally wrong to delay self-driving car technology out of fear of being criticized or because of the risk of product-liability lawsuits.

Other car companies, such as Toyota, have urged a more cautious approach because people are generally less tolerant of deaths caused by machine error than those caused by human error.

The risks associated with automated vehicles are different from the risks from cars today and could include difficult-to-find safety defects buried in software code with broad potential impacts across the industry, says David Friedman, director of cars and product policy and analysis for Consumers Union, the policy and mobilization division of Consumer Reports.  

“People just want answers to sensible questions, such as whether they’re actually safer in a self-driving car and whether their government will make sure that’s the case,” Friedman says.

Consumer Reports thinks self-driving cars have enormous potential and represent the single biggest change to cars since the invention of the motor vehicle. CR has advocated for automakers to introduce automated driving systems responsibly, both by making consumers aware of the technology’s capabilities and limitations and by ensuring that self-driving systems are used only as intended and are safe. And after safety systems are demonstrated to be effective, CR believes the systems should be standard equipment, not offered as part of an expensive luxury package.

“Today’s self-driving-car debate is not about whether the perfect should be the enemy of the good,” Friedman says. “It’s about whether or not the cars will be safer for consumers.”

CR will support any new technology that advances the needs and interests of consumers, but safety is always going to be the top priority.