robot roadblocks

The battle of innovation vs public safety continues, this time being fought on the streets of San Francisco (as opposed to The Streets of San Francisco – a Quinn Martin production). In this case, the fire chief of SF is complaining that driverless taxis from Wayne and Cruise are causing problems for fire operations. One issue is the phenomenon of “robot roadblocks” where the cars stop in traffic for no apparent reason. Sometimes they can be restarted by remote operators, but other times they require a human to lay hands on the vehicle to remedy. While there haven’t been accidents per se (which is the main metric that gets reported), there have been 39 incidents noted by the fire department, with issues of blocking access to stations, obstructing alternate routes, and even driving through taped-off areas.

While the algorithms that power the cars will improve with more training. there is a question of how quickly cities adopt the technology. Somewhat of a chick and egg issue given the need for training data, but where public safety balances with innovation is a thorny issue. In this case, two different state entities are responsible – the Department of Motor Vehicles (because they are vehicles), and the Public Utilities Commission (because they are taxis). Then you have the city officials, and of course the public. Problem is the companies do not share a lot of data with…well…anyone, so making an informed decision is challenging.

As AI moves into the mainstream, these policy questions will come fast and furious (not Fast and Furious). Figuring out what government entities are responsible for decision making is part of the puzzle, but broader questions issues liability abound. Who has the data, who can see the data, and who makes decisions based on the data are questions that need to be sorted out like yesterday…