Here's the math that Intel claims proves self-driving cars are safe

The auto industry believes self-driving cars are the future of transportation and could be a key factor in preventing accidents, but the public is wary of autonomous vehicles (AVs). So companies are looking to find new ways to convince consumers that driverless cars are a good idea.

Now, Intel says it can actually demonstrate exactly how safe AVs can be — and believes it has the math to prove it.

Amnon Shashua, who is the SVP of Intel's Autonomous Driving Group and CEO of recently acquired Mobileye, just published an academic paper that claims to provide a mathematical formula that can be applied to AVs to make sure that they won't cause accidents on the road. Intel and Mobileye also released a "layman's summary paper" on the formula, for those of us who never advanced beyond basic algebra.

Shashua and his co-authors' solution describes a "Responsibility Sensitive Safety" model, which Intel believes can be used to create autonomous vehicles that will never be responsible for a car accident. The model describes scenarios in which AVs will use pre-programmed systems to behave safely, like a "Safe Distance Formula" to handle highway driving.

Importantly, the model isn't designed to eliminate all accidents — the framers acknowledge that there will be human drivers on the road with driverless cars for "decades," and people will still make mistakes and crash their cars. We've already seen human-AV accidents during Google's (now Waymo) and Uber's on-road trials.

The point of the formula, then, is to take away any question of liability from the autonomous systems during those incidents, which is a far cry from eliminating crashes entirely. Intel believes that AVs operating with a Responsibility Sensitive Safety model will be essentially blameless, which is probably not the first thing you want to tell someone after a fender bender.

Original image has been replaced. Credit: Mashable

Unfortunately, this appears to be the type of reasoning that makes the public nervous about AVs in the first place. People are skeptical about the decision-making process behind AI systems in general, but particularly when it comes to self-driving cars.

The new Responsibility Sensitive Safety model could make AVs the most accident averse vehicles on the road, but for many, the only way they'd feel comfortable letting a robot take the wheel would be if they're guaranteed to have an absolute zero chance of a crash. Until humans stop driving, that's impossible — so maybe Intel should stick to trying to win the public over to driverless cars with LeBron James instead of math.