Waymo is 5x safer than the average driver — but married 60-year-old women in Massachusetts still edge it out
Jan 20, 2026
Key Points
- Waymo's 0.75 injury crashes per million miles beats the average driver five-fold, but a married 60-year-old college-educated woman in Massachusetts driving a luxury SUV on Tuesday morning logs 0.5, narrowly edging the robotaxi.
- Waymo outperforms median human drivers decisively but fails to match elite driver cohorts, placing it in the same zone as other AI systems: better than average, worse than the best.
- Deploying Waymo in high-risk areas like rural Mississippi would save the most lives, but market logic favors affluent tech-forward cities where safety gains are least needed.
Summary
Waymo logs 0.75 injury crashes per million miles versus 4 for human drivers broadly, making it roughly five times safer on average. That comparison flattens an important detail. When human drivers are segmented by demographics, location, and time of day, a specific cohort beats Waymo. A married 60-year-old college-educated woman driving a large luxury SUV in Massachusetts on a Tuesday morning logs closer to 0.5 injury crashes per million miles.
The data comes from NHTSA injury crash rates. Tuesday mornings rank safest—daylight, weekday traffic, sober drivers. Midnight Saturday ranks most dangerous, when people leave bars, drive faster, and are potentially impaired. Stacking the lowest-risk demographic attributes creates a synthetic cohort that Waymo does not quite match.
The opposite extreme illustrates the spread. An 18-year-old single male with a DUI driving a Dodge Challenger at midnight on a Saturday in rural Mississippi represents the riskiest driver profile, roughly 40 times more dangerous than the Massachusetts cohort.
This reframes what superhuman actually means for Waymo. The system beats the average driver decisively, which makes it a clear public safety upgrade. True superhuman performance would require beating the best humans, not just the median. Waymo currently sits in that gap between average and elite. Several other AI systems are clustering into the same zone.
A second implication emerges from the risk distribution. If the goal is maximizing lives saved, Waymo should deploy in high-risk areas first—rural Mississippi, not Boston. Market logic and customer acquisition likely favor affluent, tech-forward cities instead, which means the safety benefit is being captured where it is least needed.