The car that talks and the car that won't.

So this is interesting. Chip Motors is taking pre-orders for an electric car, and the notable thing about the launch video is that the car is presented as having a voice interface, personality, and broad social awareness:

Looking at their website I’m not sure how much the video is intended to depict the car’s actual behavior versus an imaginative conceit. I’m inclined to take it literally because my robot Sparky is doing these things right now, on my desk, and he doesn’t even have a launch hype video.¹

But this kicked off an interesting conversation on X, where John Hanacek made the following point about robots and social context, and how the Waymos handle it:

Exactly, robots need to be able to have different social contract paradigms too: civilian banter vs first responder lock in.

Seeing all the Waymo/police interactions makes my ocean lifeguard heart so sad, robots should be able to dynamically peer with professionals on scene.

— @johnhanacek

The Waymos are indeed taking another route.

The Waymos present a delightful and humane experience, but the experience is designed very much to downplay the expectation that the car understands you or wants things.

It greets you by name and speaks to you when you get in, but in a second you realize you can reply only by tapping a screen and moving through predefined steps. So the human/car interaction is like an old linear video game level, which keeps you on a narrow path. This is ironic considering the car/world interaction is much more open-ended. The car navigates a city map in real time, following the laws, avoiding other cars and pedestrians, and behaving very much like it “wants” to reach its destination.

So the design says, “I’m smart enough that you can trust your life to me understanding city traffic. But ho hum, don’t mind little old me, I’m just a car, and I’m not smart enough to understand you.”

Significantly, the Waymo design does not include a humanoid taxi driver who “drives” the car, asks about your destination, and chats about the weather. This would have produced a stronger and more alarming experience. Journalists would ask the driver how it felt about stealing jobs, about robot emancipation, etc. I see Waymos every day and they seem like creatures from Miyazaki films, large, lumbering, mechanical beasts, helpful and almost idiotically simple. Friendly, helpful, magical, harmless. For Google, this is a safe and probably an ideal result!

But it feels transitional and temporary. Waymos would be better if they incorporated more social intelligence — if they greeted you by name and then also heard your reply and could respond to questions. And also if, as John suggests, they could understand and respond to the social world outside the window, by recognizing emergency personnel and following instructions which people would naturally say rather than tap.