# Autonomous Vehicles ## Risks Good article [1] on the risks of autonomous vehicles (AV). In fact, I have recently experienced phantom braking in my 2023 Honda Rodgeline with forward collision avoidance system. It can be a very scary experience. “The difference is that while a language model may give you nonsense, a self-driving car can kill you.” The fact is self-driven cars experience about twice the rate of rear-end collisions compared to cars driven by hunans. The June 2022 accident involving a Cruise robotaxi happened when the car decided to make an aggressive left turn between two cars. Cruise cars traveled through active firefighting and rescue scenes and drove over downed power lines. Neural networks struggle to perform even basic operations when the world does not match the training data. In fact, "lack of quality data" was cited as the reason for the failed OpenAI robotics program which really was never feasible. The engineers should have anticipated model drift and other technical issues like this since these are known problems with AI models. When AI presents clear safety risks, regulatory agencies should be proactive in setting limits. Unfortunately, government bodies lack the technical competence needed to effectively oversee advanced technologies, especially AI. ## References [1]: M. L. Cummings, “WHAT SELFDRIVING CARS TELL US ABOUT AI RISKS”, IEEE Spectrum magazine, October 2023. [2]: [Exploring the Potential and Risks of Autonomous Vehicles](https://friedmind.hashnode.dev/exploring-the-potential-and-risks-of-autonomous-vehicles) [3]: M. L. Cummings, "[What Self-Driving Tells Us About AI Risks](https://spectrum.ieee.org/self-driving-cars-2662494269)," IEEE Spectrum, vol. 60, no. 10, pp. 30-35, October 2023, doi: 10.1109/MSPEC.2023.10271347.