From a 0% base two years ago to 10% of ride shares in some US cities, the A.I. behind self-driving cars is maturing and coming down dramatically in price. Here's how it will overhaul cities and whole economies:

WHERE WE ARE TODAY

Waymo is running fully driverless ride-hailing across San Francisco, Phoenix, Los Angeles, Austin and Atlanta, logging hundreds of thousands of paid rides weekly.
Expansion is accelerating — half a dozen more US cities are in the pipeline, plus London as Waymo's first international market.
In China, Baidu's Apollo Go, Pony.ai and AutoX already operate large-scale services across Beijing, Shanghai, Wuhan and Shenzhen, with Apollo Go delivering millions of rides per quarter.

THE ECONOMIC TRANSFORMATION

The math is simple: no driver means labor cost per mile drops dramatically, and vehicles can operate far more hours than privately owned cars that mostly sit idle.
The average US household spends ~15% of its budget on vehicle ownership so "subscribe to mobility" will be very tempting for city-dwellers once per-mile prices fall at scale.
US downtowns dedicate 20-30% of land to parking. Reduced car ownership could unlock surface lots for housing, parks and offices — and convert curbside parking to wider sidewalks and bike lanes.

SAFETY

For those of you concerned about autonomous-vehicle safety, Waymo's safety data show serious-injury crashes roughly ten times lower than human benchmarks.
Want third-party verification on that 10x safety improvement? Swiss Re found ~90% reduction in bodily-injury and property-damage claims for autonomous vehicles relative to human drivers. (Perhaps in the not-too-distant future, insurance premiums will become exorbitant for human drivers, pushing more and more people to go autonomous 🤔)

RISKS

Cheap, effortless rides are a recipe for gridlock without smart policy. NYC's congestion charge cut incoming traffic ~10% in its first months, for example — dynamic "robotaxi fees" like this will be essential.
Labor impact is significant: ~500K taxi/shuttle jobs, ~500K bus-driver roles, and ~3M truck-driver jobs in the US alone. Reskilling pathways and transition plans are critical.

THE GENERAL LESSON: Where in your field is A.I. currently handling only a small percentage of a particular workflow autonomously, but is poised to take over most or all of the workflow as LLM capabilities improves and costs continue plummet (they're currently falling at 100x per year)? Huge opportunities lie there for you (and your organization).

The SuperDataScience podcast is available on all major podcasting platforms, YouTube, and at SuperDataScience.com.