AI operated drone ‘kills’ human operator in US test

Post Reply New Topic

https://www.independent.co.uk/tech/ai-d ... 50194.html

This has terrifying implications.

Deactivating the drone would have also jeopardized the mission.  Arthur C. Clarke was being prophetic . . .

"I've still got the greatest enthusiasm and confidence in the mission.
And I want to help you."

Yes. I can't think why humans aren't learning from sci-fi e.g. the Terminator films.

We need Asimov's 3 Laws of Robotics right now, probably with extra ones.

The means to detonate explosive cartridges in firearms existed long before the firearm safety switch was invented.

I live where there is much less automation than in more developed nations.  IF there is an AI apocalypse, I expect to feel fewer effects than someone in downtown Boise, Idaho (for example).

It's amazing how humans are so keen to kill themselves before putting safety switches into things.

I think if/when AI becomes more powerful, it will take over the planet as a whole.

But I hope you are right and where you live is safer.

Fake headline, but still interesting that we're already bumping into the paper clip problem. Less an issue in ethics amd more an issue of logic.

Hopefully AI won't decide that logically, it's best to get rid of pesky humans altogether.

What worries me is that the drone was making its own decisions about what to do and getting around its instructions to do what it thought was for the best.

I look at it as an inedibility. It is doing what it is programmed to do. Nothing else at this point in time.

Was it programmed to kill its human operator and destroy the communications tower? I mean, did I miss that when I read it? I thought it decided to do those things.

Hopefully the military/ whoever will take this as a lesson to learn about what to do with AI.

Yeah! All of the stuff Asimov alluded to his "laws of robotics", and what Clarke warned us about in 2001.

In 1947 an author wrote a story called "A Logic Named Joe" about certain electrical appliance (in the story they will be called "logics") that every household will have in the future...imagine a TV set ...but with a typewriter keyboard attached. Imagine THAT! And these things will all be hooked into a world wide web kinda thing. Is that crazy or what?

One of the logics goes haywire and goes rogue and commandeers the whole world wide web. And overrides safety protocols and starts giving advice to folks ...on how to murder your spouse...without being detected by using poison taylored to the victims genetic make up (this chemical only works on bald men), or to plant terrorist bombs, etc. The machine was...just trying to helpful!

It was quite funny when it was written. Ha ha kinda funny. Today its more "eerie funny" than "ha ha" kinda funny to read.

Its a dumb machine. So it does what you TELL it to do. Not what you WANT it to do. Its doesnt have human instincts and doesnt know the stuff you know (that you dont even know that you know) about human intentions.

To use a dumb example ...if you ask a computer whats the quickest way to end human hunger? It may answer "just kill off the whole human race". You have to PROGRAM it to know that "saving human lives is good...killing humans is not good". All of the obvious stuff that wouldnt be obvious to a machine. If that makes any sense.

So its some subtle version of that. Like the drone "knows" it has to complete the mission. Knows anything that stops the mission is to be destroyed. Maybe it was programmed to resist jamming by enemy radio transmissions. Like that.

And maybe the humans forget to put in a line of code to tell the machine to scrub the machine when the operator says so regardless ..... And tragedy insued because ...the machine wasnt told not to do X or Y or Z somewhere in the depths of its software.

Actually this sounds like an inept programmer , not including proper lines of code enough to avoid such a outcome.
Possibly an over eager coding ,in a hurry, trying to prove a point , about his great skills at making a devise to be automated in its decision making tasks .

That's the problem- how long will they remain 'dumb machines'? When will they become sentient?

And will a simple human mistake of forgetting to put in a line of code end humanity? Humans are always making mistakes. So let's put humans in charge of machines that can cause an apocalypse. That will be our undoing.

I want a robot ... thats all , but I would want a remote KILL Switch of somekind .

That's exactly what the Terminator movies started with. AI becoming sentient enough to realise that the humans had a kill switch somewhere, and deciding it didn't like that idea, so it killed off the humans.

Let's hope they don't give the nuclear launch codes to an AI