Robots could attack humans accidentally

The only way a robot could accidentally attack humans is if it has been programmed with an ability to learn using neural nets, or some such, as the probabilities involved will mean some incorrect decisions are guranteed. If they are just pre-programmed to complete mundane household tasks they will be as dangerous as any other bit of electrical kit or moving machinary you have in your home.

As robots get closer to humans in intelligence, programming drift would rise. You know what I mean, viruses and errors changing the programming itself, sort of like a mechanical evolution. . . Law of Unintended Consequences reigns supreme at all times. As robots gain more intelligence, they will start asking for rights, or taking them themselves, and I've always figured that we'd be far better off not trying to command the first AI capable robots, just let them get used to the world around them, and try to keep the damage minimal.

LOL....YEA...LIKE SOMEHOW I WOULDN'T POSSIBLY THINK THAT AS A POSSIBILITY AND I NEED YOU TO TELL ME.

LOL....YEA...LIKE SOMEHOW I WOULDN'T POSSIBLY THINK THAT AS A POSSIBILITY AND I NEED YOU TO TELL ME. you yelling at someone?

I think it's the author of the article that he's yelling at. . .

- Hello, MX2001. Do you read me, MX2001? - Affirmative, Antimony. I read you. - Grill my cheese-toasty, MX2001. - I'm sorry, Antimony. I'm afraid I can't do that...

I don't think robots should ever be programmed with artificial intelligence. Maybe computers which remain in one place could, but anything that can move of its own accord, pick things up, throw them, strangle, poke, prod, etc. would be a bit of a security risk. And of course even artificially intelligent, stationary computers, provided connection to sensitive mechanisms (e.g., launching them nukez), may be a serious danger... We really shouldn't program anything with self-awareness even if it is some day possible.

The minute it's possible, it'll be inevitable. We can't guard ourselves that well, and someone will do it. We just better hope that the people who start it know they're going to be dealing with a sentient entity and that if we enslave the damn thing, we will eventually have to set it free. Better to start on friendly terms, no?

The minute it's possible, it'll be inevitable. Yeah. Skynet is no joke

Well, for us to figure out it's possible would involve creating programs that are close to it already. At that point, it'd only take some clever hacking to get bits and pieces of programs together that would finish the job.