Elon Musk and Stephen Hawking Join Call for Ban on Artificially Intelligent Weapons

I'd worry more about the digital aspects of A.I. weapons than a physical aspect of an A.I. weapon.

You know, with the way the world is wired now, a truly A.I. weaponized virus would basically hold our civilization hostage.

humans need to key in characters on a keyboard to combat code, an A.I. would simply need to think it into reality. Who you think will win against a A.I. hacker.

Now whether such a ban can be implemented is the greater question and imo impossible.

originally posted by: pl3bscheese
The way I've looked at grey-area moral issues is has to include the line, "if not us, who">? If not Western forces, then it will happen with other countries. If not by the military, then by curious mad scientists. This stuff is inevitable. This public speaking brownie points can play out as it will, but nobody should be fooled into thinking it'll actually be kept under wraps.

We've done a pretty good job with nuclear non proliferation so far. Not creating autonomous weapons isn't all that much more difficult.

To be honest, some of our deadman nuclear systems from the cold war aren't all that different from what we're talking about here. The only difference is we're talking about deploying them on battlefields rather than a single binary choice of "is our nation about to get destroyed by nukes"?

Building these systems is very dangerous, but someone somewhere is going to do so. The best thing we can do is delay that happening until a countermeasure is already invented.

originally posted by: Isurrender73
a reply to: neoholographic

Hasn't anyone watched Star Wars(the mindless controlled by one evil force), or Terminator (the machines taking over)?

AI weaponry is a bad idea. Even it the AI doesn't try to take over, you would be putting unlimited military power in the hands of very few individuals.

I would counter with Deus Ex, Ghost in the Shell, and Star Trek if we're going to take random fictional ideas and extrapolate reality from them.

In Star Trek specifically they had an interesting take on it. They had the capability to create self aware AI but for the most part considered it inhumane to do so. Holodeck personalities for example could become sentient and therefore actual life forms but when their program stopped being run they would essentially die. As a result AI's were deliberately kept dumb, and the few that did become self aware anyways were granted personhood with modified programs that saved all of their memory, along with control over turning their own program on/off as if they could wake.

I'm fairly sure there will be battle-management AI's that do pretty much everything except make the actual decision to kill. In the future a UAV will fly itself on a patrol and when it finds a target, it'll have to get a human to "approve" the strike.

That's probably how it'll work.

originally posted by: neoholographic

originally posted by: Rikku
that article is kinda nonsense.
but if stephen hawkings has been quoted, it must be serious.

I don't think it's nonsense in fact it's not just Hawking. It was a letter signed by 1,000 experts.

Musk, Wozniak and Hawking urge ban on warfare AI and autonomous weapons

More than 1,000 experts and leading robotics researchers sign open letter warning of military artificial intelligence arms race

www.theguardian.com...

I think they're worried about 2 major things. The fact that this area is advancing rapidly and they're asking if you create an intelligence that's smarter than you, how can you control it. I think these questions need to be asked.

That's like asking how yiu control your child if they are as smart as hawking

The answer is easy

You don't

You hope you child is raised right and makes good decisions

The fact we are even thinking abiut controlling it is as absurd as controlling another person

You don't have the right to control an ai just as you have no right to control another person

If you create life then its the right of that life to live as they wish

Period

a reply to: Aazadan

Actually, that's not the correct way of looking at it. Look at how they were produced in the first place. It was a race. That's the way to look at this. No matter what ethics are pushed, it will be a race to have the first military with this edge.

Nuclear non-proliferation has not been successful in half a century as far as keeping humanity from destroying itself. We have multiple times over the amount to do so still. What it did was switch focus elsewhere, namely space.

This is one of many things I apply the same logic of "if not us, who else"... because that initial curiosity and edge will be sought be some entity no matter what. We were unable to put the genie back in the bottle with nukes, no reason to think this will be any different.

originally posted by: MystikMushroom
Perhaps all the flight data from drone pilot input is fed into an AI so it can learn to fly like how a human would?
Just speculation.

Been there, done that, have the tshirt.
Back in 2007. Although, you can replace the "data from drone pilot" with internal and external sensor data back fed into the virtual environment the bird was moving in. Except it was learning how to fly like a bird, not learning like a human would operate a planes controls but the principe stays the same.

My only fear is that we, as fallible human beings, would transfer elements of our infallibility into the artificial intelligence that we're creating. It may very well, do things we would not expect it to do and react in ways that are eerily similar to us. In fact, who says that we'll be the enemy that AI's decide must be eliminated? What if the AI's actually engage in literal and political warfare with each other? What if they simply disable our defenses in order to fight what is, to them, a more important war against other AI's.

The best AI implementations will probably be expensive.

That is usually synonymous with central government run projects.

Theoretically it might not be a threat to a site like ATS since advertizing revenue is generated from click through.

Content will suffer for many, but there are game cheats that will be discovered and shared.

The UK is a small country with limited budget, we saw what happened to Gary McKinnon the British hacker who lives with his mum and chastised the US military sites after 911. Philosophically speaking the invention of the printing press was a huge step forward for AI and so was the Babbage engine. This looks like a topic that has a lot of scaling potential...

DBL Post

A.I based weapons like turrets would have its pro's and con, due to how they are autonomous and vigilant, but the problem it would have would be telling one human from the other, or adapting or using "Critical Thinking" to a new situation. Especially something identifying like race not a big issue as it use to be, so it would have a real hard time distinguishing friend from foe unless their was something that could identify with like a tag, or facial recognition. God forbid if you had big enough pimple or bad enough cold on the same day and it didn't recognize you, the contrapation would be so dam trigger happy.

A.I being the central Nervous centre, now whos the genius that destroyed humanity?

originally posted by: Specimen
A.I based weapons like turrets would have its pro's and con, due to how they are autonomous and vigilant, but the problem it would have would be telling one human from the other, or adapting or using "Critical Thinking" to a new situation. Especially something identifying like race not a big issue as it use to be, so it would have a real hard time distinguishing friend from foe unless their was something that could identify with like a tag, or facial recognition. God forbid if you had big enough pimple or bad enough cold on the same day and it didn't recognize you, the contrapation would be so dam trigger happy.

A.I being the central Nervous centre, now whos the genius that destroyed humanity?

Identifying a friend is relatively easy. You just need an identifying mark that's hard to reproduce on your soldiers, ideally on a uniform, for example an optical camouflage pattern that the camera recognizes as an ally. An enemy is more difficult because neither neutral parties or enemies have a distinguishing feature to separate one from the other.

originally posted by: MystikMushroom
Who's to say the contractors working for black military projects haven't been developing and creating AI's already?

Perhaps all the flight data from drone pilot input is fed into an AI so it can learn to fly like how a human would?

Just speculation.

People are saying this Jade Helm exercise is actually computer controlled. It was done to test the supercomputer in the real world.


Personally I am in the Ray Kurzweil camp.....It is coming and the future will be really cool.....Virtual reality,Living to be 150 years old,self driving cars.........Saved the biggest one for last......3D printing. Additive manufacturing will change everything. Darn I left out graphene.