[falcolas]

"Manifesto" talk aside, this is the same discussion about automation styles: Ironman vs. Ultron.

I'm a huge fan of Ironman automation method - augment, not replace the humans - since it takes the boring and error prone parts away, and utilizes humans for the edge cases. There's a lot of potential for huge wins for low costs. And, absolutes aside, that's in line with this article.

The alternative Ultron method - take the humans out entirely - is simply too expensive and error prone. It's absurdly common to watch entire automated assembly lines grind to a halt because of corner cases that have to be fixed by humans. We're seeing the same thing with automated vehicles - when corner cases are hit, they pull over and signal for human intervention.

So, I guess I agree with the author, though in a less stylish way. Please make me a better driver; don't remove me from the seat entirely. Just imagine that fancy VR simulation they showed from the Waymo cars made available as HUD data for a regular driver. How much would that alone help with safety on the road?

[maxerickson]

Where does the Vision fit into it all?

[falcolas]

Probably somewhere around the time we can directly connect our brains to computers, Ghost in the Shell style.

Which, by the way, I'm all for. Keyboards, steering wheels, touch screens... they are all just too limited for proper human-computer interaction.