killer robots

Humans make more mistakes than computers.

Humans program the computers...

Programs can be peer-reviewed.

I love the clip, but it's clearly sample biased.

Sure, some robots might kill people every once in a while, but people kill each other every ten seconds.

It's like the panic over people killed in self-driving cars.

Which do you think kills more people, self-driving-cars or people-driving-cars?  Wild guess?

I'm not a luddite, 3R! 

Accidents - 'blue on blue' and 'collateral damage' occur in wars and i think its highly probable that robots will make fewer errors than humans.  It's not ust about them goib wrong though... it's also about when they work according to plan!  Rhe big questio being what 'the plan' should be!

My guess is that the tech isn't quite upto producing an effective fully autonous weapon - yet.   We're decades at least off creating robot infantry.  A more realistic scenario is a drone being programmed to seek and destroy a particular enemy asset such as radar installation.  But suppose the enemy has, unknown to us, deploys a human shield?  Do we need a human veto?

This should make you feel better. [LINK]

Nice. [LINK]

Nice, like dem flared and finned bellbottom leggings.

I was watching "This Giant Beast" [LINK] last night and they had a really good episode on AI.

One of the test problems they solved with "Unanimous AI" [LINK] is the self-driving-car trolley problem.

If you MUST kill one of the following groups, which do you choose?

1. a baby

2. a young girl

3. a young boy

4. a pregnant woman

5. two male doctors

6. two female doctors

It's been worked out already....

Everything anybody will ever need to know can be learned by studying the '70's.

Currently; computers can only make mistakes when humans make mistakes.

I imagine that one day fairly soon it will be possible for computers/A.I. devices to make mistakes all by themselves.