How has your academic background in philosophy and neuroscience informed your work in the automobility industry?
After growing up in Austria, I arrived in the U.S. in 1988 and became interested in conducting research on neural network modeling after discovering that the artificial intelligence industry was not making much progress at the time. When I was employed at McKinsey, I was able to complement this technical knowledge with my experience learning about how big industrial companies conduct transportation logistics and field services. In addition, my background in philosophy also helped me recognize the technical and ethical limits of technological tools like artificial intelligence and neural networks.
In 2015, I founded Nauto intending to address some of the challenges that came with the shift toward automobility that had begun to take place in the industry — a shift that promised to fundamentally alter how cities were built and operated. The challenge was designing an autonomous vehicle that could run on artificial intelligence software with a 99.9% reliability index (higher than that of the average human driver). Yet for AI to reach this degree of reliability, it will need to have processed at least 200 billion miles of “driving experience” to account for improbable accidents. And it will take approximately another decade to obtain the necessary data for this kind of technology to work at scale for general purposes.
So, Nauto focused on developing technologies that aid, rather than replacing human drivers by providing them with basic automated safety mechanisms that preempt risks you or I might take unconsciously while driving, such as falling asleep at the wheel, getting distracted by our smartphones, or failing to detect safety hazards within a 360-degree scope.
Could you elaborate on the kinds of technological solutions Nauto offers to address these safety hazards?
The technology we employ is like that used in autonomous vehicles. For instance, we develop computer vision to identify objects — say, cars approaching a neighboring lane — and assess the risks they pose to a driver. Yet the difference is that our company produces computer and sensor packages that cost about $500, as compared to the price of a $200,000 automated car.
And Nauto is unique in that it not only implements forward collision alarms (i.e., to avoid risks that exist outside of the car) but also protects against human errors that occur inside of the car (and make up 96% of the causes of vehicle accidents). In other words, we use the same computer vision technology to also track the driver’s field of vision, attention to the road, and real-time awareness of external risks.
So, the breakthrough in Nauto’s work is that it surfaces the unconscious risks we take as drivers and thereby enables us to create for the driver what I informally call “Oh shit!” moments in which they can actively recognize their dangerous behaviors behind the wheel. Instead of using a conventional beeper alarm, the Nauto solutions uses voice commands like “pull over to use your phone” or “leave more following distance from the vehicle ahead” to intervene in what we have been describing as risky driver behaviors. Essentially, our technology employs artificial intelligence to alert and train drivers to develop safe habits, ultimately reducing the number of collisions by about 60 to 70%.
What next steps is Nauto going to take to have this technology reach a broader consumer base?
Our objective is to work with multiple automakers on integrating our deep learning algorithms into a variety of cars and thus avoid having customers retrofit a separate hardware system for safety. We have already been collaborating with General Motors to install our technology in their electric BrightDrop vehicle. We also aim to continue informing consumers on how our artificial intelligence processors work to ultimately reduce the average insurance risk of $3,000 to $7,000 per commercial driver by about two thirds, while reassuring them that they are not being “surveilled” because it’s a real time AI copilot helping then detect and avoid risks and collisions and not a supervisor watching them.
