In a fascinating evolution of autonomous driving technology, Waymo – Alphabet's self-driving unit – has deliberately reprogrammed its robotaxis to behave less like perfect rule-followers and more like experienced, assertive human drivers.
The shift, which Waymo describes as "confidently assertive," has transformed vehicles once criticized for being overly cautious into ones that now accelerate promptly on green lights, perform rolling stops at quiet intersections, and even honk when cut off. Some San Francisco riders now liken the experience to riding with "an aggressive New York taxi driver."
This change didn't happen by accident. Early Waymo vehicles were programmed for extreme politeness: they yielded excessively at four-way stops, waited patiently behind double-parked cars, and came to complete halts even when no one was around. While safe on paper, this hyper-cautious style created real-world problems.
The cars frequently got stuck, disrupted traffic flow, and frustrated passengers in a hurry – locals often avoided them when time was short. Scaling a commercial robotaxi service in dense, chaotic cities like San Francisco demanded a rethink.
Waymo's engineers discovered an unexpected truth: pure rule adherence wasn't always the safest or most efficient approach. By studying millions of miles of data, they found that blending into human traffic patterns – anticipating moves, claiming space confidently, and avoiding unnecessary hesitation – made the vehicles more predictable to others on the road.
As one Waymo product director explained, assertiveness helps the cars "do things that you expect other humans on the road to do," reducing confusion and potential conflicts.
The results speak volumes. As of mid-2025, Waymo's fully driverless fleet has logged over 100 million miles on public roads, including nearly 100 million rider-only miles across cities like Phoenix, San Francisco, Los Angeles, and Austin. Independent analyses and peer-reviewed studies show dramatic safety gains: up to 91% fewer crashes involving airbag deployment, 96% fewer injury-causing intersection incidents, and an overall 85% reduction in suspected serious injuries compared to human drivers in the same areas.
Vulnerable road users – pedestrians, cyclists, and motorcyclists – benefit most, with injury crashes down 82-92% depending on the category.
Of course, the transition hasn't been flawless. Isolated incidents have made headlines: an illegal U-turn that led to a police stop, a low-speed collision with a neighborhood cat, and occasional close calls where vehicles inch forward as pedestrians clear crosswalks.
Critics worry that mimicking human impatience could erode the inherent safety edge of machines that never get tired, drunk, or distracted. Yet Waymo insists these edge cases are rare and actively reviewed, with software updates refining behavior continuously.
The irony is striking. For years, autonomous vehicles were sold as superior precisely because they wouldn't emulate human flaws – no road rage, no texting, no speeding. Now, Waymo's data suggests that selectively adopting human-like assertiveness, while retaining perfect vigilance and reaction times, yields even better outcomes. In bustling urban environments, being too polite can create hazards; flowing with traffic, even if it means bending minor rules like a quick rolling stop on an empty street, often proves safer overall.
Also read:
- Anthropic Engages Top IPO Law Firm Wilson Sonsini, Signaling Potential 2026 Public Debut in Race with OpenAI
- YouTube's 2025 Recap: Your Personal Highlight Reel of Binge Habits, Now Shareable Worldwide
- Netflix Quietly Kills Phone-to-TV Casting for Most Users – And Almost No One Noticed
Waymo's experience highlights a broader lesson in AI development: the path to superhuman performance sometimes runs through careful imitation of skilled humans. As the company expands its fleet – now numbering thousands of vehicles and serving tens of thousands of weekly rides – this "humanized" approach could accelerate the mainstream adoption of robotaxis. Safer roads, fewer fatalities (Waymo reports zero in over 100 million miles), and smoother traffic might just come from machines that learned the best – and occasionally the pushiest – lessons from us imperfect drivers.
Author: Slava Vasipenok
Founder and CEO of QUASA (quasa.io) — the world's first remote work platform with payments in cryptocurrency.
Innovative entrepreneur with over 20 years of experience in IT, fintech, and blockchain. Specializes in decentralized solutions for freelancing, helping to overcome the barriers of traditional finance, especially in developing regions.
