[pasta]

(current) AI is not smart it's a method to have a reasonable predictable outcome from a wide range of inputs. But it is still a linear program.

Let say we program a self driving car to estimate the impact effect of a collision with an object. Then we 'learn' it that a crash with a soft object will be better for the human in the car.

So we think the car is smart because it can detect soft and hard objects. But in case of an unavoidable crash it steers into a group of people instead of a parked car...

The Pentagon (DARPA) is now investing in AI that learns from previous experiences (without feeding it a dataset over and over again). I guess other companies are working on this as well. That will create scary AI because then the program will be altered all the time making it 'smart'.

[b_tterc_p]

Pet peeve here. We don’t need to train AI to decide what to crash into. We need to teach it to try and avoid crashing, and doing so foremost using the brakes. It’s ok if cars crash sometimes. It’s not ok if cars spazz out and veer onto the sidewalk.

We definitely don’t want the car deciding prematurely that it can’t possibly avoid a crash and start choosing what to hit (or even aim for). That will lead to truly dumb Ai.

1) minimize likelihood of crash 2) minimize speed of likely crashes