NO ONE IS DRIVING THIS TAXI. WHAT POSSIBLY COULD GO WRONG?

I won’t forget the first time I took a ride in a car without anyone sitting in the driver’s seat.

It happened one night last September when a Chevy Bolt named Peaches picked me up outside a San Francisco bar. Our ensuing half-hour ride together produced, at first, a titillating display of technology’s promise. Then an unexpected twist made me worry that the encounter had turned into a mistake I would regret.

Peaches and I were getting along great for most of our time together as the car deftly navigated through hilly San Francisco streets similar to those that Steve McQueen careened through during a famous chase scene in the 1968 film “Bullitt.” Unlike McQueen, Peaches never exceeded 30 miles per hour (48 kilometers per hour) because of restrictions imposed by state regulators on a ride-hailing service operated by Cruise, a General Motors subsidiary, since it won approval to transport fare-paying passengers last June.

It was all going so smoothly that I was starting to buy into the vision of Cruise and Waymo, a self-driving car pioneer spun off from a Google project that is also trying launch a ride-hailing service in San Francisco.

The theory fueling the ambition is that driverless cars will be safer than vehicles operated by frequently distracted, occasionally intoxicated humans — and, in the case of robotaxis, be less expensive to ride in than automobiles that require a human behind the wheel.

The concept does sound good. And the technology to pull it off is advancing steadily, just like other artificial intelligence applications such as chatbots that can write college-level essays and produce impressive pieces of art within seconds.