In August 2023, the California Public Utilities Commission approved Waymo (a subsidiary of Google) and Cruise (owned by General Motors) to start charging customers for rides in driverless taxis. By the first quarter of 2025, these autonomous vehicles had collectively driven 80 million kilometers. A 2026 study by the Insurance Institute for Highway Safety (IIHS) found that Waymo’s robotaxis had a 68% lower rate of bodily injury crashes compared to human drivers per mile driven. The study covered major U.S. cities like San Francisco, Phoenix, Los Angeles, and Austin. This data suggests that autonomous vehicles may be significantly safer than human-driven cars in certain conditions. Swiss Re, a top global reinsurer, reported that Waymo's autonomous vehicles had 88% fewer property damage claims and 92% fewer bodily injury claims compared to human drivers. They also saw a 92% drop in severe or fatal accidents and a 96% reduction in incidents involving pedestrians. According to the National Highway Traffic Safety Administration (NHTSA), over 90% of fatal accidents are caused by human error, including speeding, alcohol use, and distractions. Removing these factors could save thousands of lives annually worldwide. A 2024 study published in Nature Communications confirmed that autonomous vehicles are significantly safer in routine driving scenarios such as maintaining lanes or driving straight. Researchers found that self-driving cars crash less frequently and with fewer injuries than human drivers. However, some experts warn that comparing autonomous vehicles to human drivers may be misleading. Robotaxis often operate in carefully mapped areas, unlike the unpredictable and varied conditions human drivers face. Autonomous vehicles are designed to be safer due to their lack of human weaknesses, such as fatigue, distraction, or alcohol influence. They use a 360-degree vision system combining high-definition cameras, radar, and LiDAR sensors to create a 3D map of their surroundings. These systems react faster than humans, detecting dangers like a car running a red light and braking instantly. They are also programmed to follow traffic laws precisely, maintaining safe distances and smooth acceleration and braking. Despite these advantages, challenges remain. Autonomous vehicles are more likely to have accidents during dawn or dusk due to changing light conditions and are nearly twice as likely to have incidents during turns or complex intersections. Weather conditions like fog or heavy rain can interfere with sensors or overwhelm perception systems. Additionally, these vehicles are vulnerable to cyberattacks, as their interconnected systems could be targeted for hacking, leading to data theft or vehicle takeovers. Researchers have also pointed out methodological issues in comparing autonomous vehicles to human drivers. Philip Koopman, a researcher, argues that comparing a new robotaxi in a controlled environment to the average human driver—many of whom are young, inexperienced, or driving older vehicles—is misleading. The Johns Hopkins School of Public Health notes that autonomous vehicles have driven far fewer miles than humans, and data collection across the U.S. lacks standardization. Some road safety advocates suggest that countries like Sweden and the Netherlands have greatly reduced road deaths through infrastructure improvements such as protected bike lanes and lower urban speed limits, without relying on autonomous vehicles. However, industry observers argue that the promise of fully safe autonomous driving is being tested by real-world incidents. For example, in Wuhan, China, over 100 Apollo Go robotaxis from Baidu were caught stopping simultaneously in the middle of an intersection and on highways, causing traffic jams and minor collisions. Passengers were trapped inside and unable to open the doors or call for help. On the other hand, real-world data and video evidence show that autonomous systems can outperform humans in dangerous situations. A video shared by Waymo co-director Dmitri Dolgov shows a robotaxi on the 10 Freeway in Los Angeles avoiding a collision with a human-driven car that had lost control. The autonomous vehicle detected the danger early and executed a precise maneuver without panic. Industry observers argue that the risks of human error justify the push for autonomous vehicles, even if they are not yet perfect. “Anyone who opposes autonomous cars encourages tens of thousands of deaths,” says Eric Newcomer, who believes human drivers are responsible for too many fatalities each year. He argues that while perfection should not delay progress, the removal of dangerous human drivers from the road is urgently needed.