TO THE EDITOR:
Elaine Herzberg’s death illustrates that we should reconsider artificial intelligence in cars. Dubious algorithm software is a LONG WAY from SAFELY maneuvering down the road. AI is no more and no less than what the people who write programs believe is the correct response to a situation.
It is not capable of individual thought, ergo it will apply the “solution” it has been given that most closely meets the desired result.
This accident and one involving a motorcycle in California appear to have been preventable! In California: Most of us practice extra care near a cycle; the AI vehicle was not programmed to do so. In Arizona: Three seconds to slow down and change lanes could have avoided the pedestrian. The onboard monitor was not paying attention, and the AI was not programmed to handle this situation.
They call it “artificial” because it is not as good as the human brain in assessing a new situation and reacting.
When desktop software stops working, “experts” advise a reboot. If desktop software has been around decades and STILL does that, why is beta-tested AI adequate to let loose on public roadways?
No one of Steve Jobs’ caliber is supervising this effort. He tolerated nothing, expected everything and forgave nothing. A bit over the top, perhaps, but nothing has come close to his obsession for perfection. This approach is required to solve AI. These accidents show that the AI vehicle behaves like a distracted driver. We do not need more of those!
A.M. deLANGE, Reisterstown, Md. The writer is a former oil company engineer who now is an independent provider of vehicle servicing.