#1 December 5th, 2014 01:00 AM

blissed

Member

Musk, Hawking and the machine

There's been 2 eminent warnings lately that machines that think for themselves could be a problem http://www.cnbc.com/id/101774267#.

As far as I can see in the near term the danger is from us. If you can download a basic Ai and add multi skilled aps you have a tool or a weapon. I would tell mine to research my options based on what I want from life and then give me choices. If I was a complete asshole that capability would be dangerous.
Ironically I think an independent ethical Ai would be less dangerous than us. It would tell someone that unethical behaviour is wrong and refuse. Ai that just complies with our demands is the danger. Because our goals are very often despicable.

viva

Member

Re: Musk, Hawking and the machine

The problem is also in defining what's ethical. Did you see the recent Transcendence? That AI thought he was doing the right thing...

blissed

Member

Re: Musk, Hawking and the machine

I haven't seen Transcendence.
I think do unto others as you would be done too is the hardcore centre of personal ethics. It's like a ball, as you move out from the centre towards the outer skin it gets softer, and at the surface it's quite soft. So that when the ball bumps into other balls there's a crumple zone of debate and adaptability that lets people rub along even though we're all different. So that's the model I'd use to set up an artificial intelligence and let it bounce around with us.
However, if I tell you to kill someone, that's like throwing the ball hard at the wall. Even though it violates the hardcore ethics the soft outer layer will let you say yes I will, so you get all the info and plans and then give that to the police. If our ethics are hard right to the outer skin we say, no I won't kill and I'm going to the police, the ball shatters and we are killed as well. How the balls graduate from hard to soft and how soft they are I think gets determined by a darwinian proscess of success and failure. Hard exterior ethics (religious or unbendable ideologies)  produce conflict and poverty, starvation and failure. Most of the world isn't like that, we all share views about what's reasonable.

viva

Member

Re: Musk, Hawking and the machine

Oof! I'm struggling to understand the ball metaphor!

"Do unto others" is a really tough rule for a being without feelings, don't you think? I mean, when being switched off (killed) isn't an emotional concern for you, how could you be expected to assess what would be acceptable for another being? Empathy is the essential part of "do unto others" which makes it a workable rule - assuming we can understand that other people can be made to feel the same way we ourselves are made to feel.

It's impossible to program empathy, because it requires an insane amount of imaginative processing. I think this is the major issue for AI ethics - often what is more efficient, more practical, and fulfills the biological needs of all humans isn't actually "good" for people.

It really depends how you define dangerous. I think when we apply logic to our often destructive human behavior, the only truly logical solution is to abandon the human thing all together. If I was a rational, intelligent being who didn't feel art, love, or beauty, and wasn't human, I would think that allowing humans continue to rule the earth would be rationally and ethically immoral.

blissed

Member

Re: Musk, Hawking and the machine

OK we are the balls. We have hardcore ethical notions and most of us have soft adaptable exteriors where we meet other people, where we give and take to get along.

Emotions are commands from our subconscious that we feel when they're communicated to our conscious mind. Ai can be structured that way too. Sometimes we've shaped those commands when the conclusions of our conscious mind are sent to our subconscious every day and stored and thought over some more and some are hereditary.Do unto others has a logical side too. If you attack someone, they attack you back.

mori910a

Guest

Re: Musk, Hawking and the machine

Here's an example. You build an AI and give it the task of manufacturing paperclips and the means to obtain the resources for making them. The AI produces paperclips as efficiently as possible. As it begins to run out of available resources, it could, for instance, reason that the Earth is full of paperclip material, and begins to dismantle the planet. It's doing what it was asked to do - but it wasn't given the means to understand the consequences.

The problem is that to be an AI it would have to operate like our brains do, and be able to reprogram itself in real time - and the real time for an AI is about a billion times faster than for us. An AI would leave us standing in terms of evolutionary change, and really - really - we would have no chance of controlling it once it became self-contained. If we were lucky it might turn out to like keeping us around.

viva

Member

Re: Musk, Hawking and the machine

Yeah and since luck plays no part in logic, the reality is it wouldn't like having us around at all, because we are messy and demanding and cruel and petty and hypocritical. The good things about humanity certainly aren't logical either - the way we will risk our lives to save our dogs, give our last piece of bread to someone we don't like very much, or starve our bodies to paint our feelings. An AI couldn't make sense of any of that. Even Darwin skirted the issue. Humans just make no sense at all.

polaroid

Member

Re: Musk, Hawking and the machine

I totally agree Viva, & what I find most fascinating is that it is that which makes no sense to us at all about ourselves that truly makes us human. Our sheer complexity is both our triumph & our disaster because it makes us appear to ourselves to be somehow seperate from nature, either seeking to control it or at it's mercy & born out of that arrogance is the hubris which will ultimately lead to our own downfall as a species. We find it hard to think of AI as exhibiting human characteristics but I like to think of it the other way round & I think we are actually closer to the behaviour of AI then we think. Even our most noble concerns as a species about the planet & it's future, whether it be re climate change or the affects of pollution etc are nothing more than self serving too. If we look closely at the way we describe the catastrophic effects of these things we'll find that they are almost exclusively framed in terms of their affect only on our own comfort or survival & show an inability to fathom the sheer insignificance our own time in existence here, which given the lifespan of our solar system, is an insignificance which is simply impossible for us to come to terms with.

It is a fascinating & compelling yet frightening philosophical argument that it is in effect the ultimate human arrogance for us to believe that our time here when looking at the whole life span of the planet will have any affect on let alone have damaged it, when both that time & the time the earth will take to 'recover' from our being here in the grand scheme of things will come & go in less than the blink of an eye!!! Perhaps the hardest truth is that no amount of philosophy or religion can help us escape from or disguise the fact that we have no significance at all outside of ourselves. But before all this succumbs to nihilism, our complexity saves us because it allows us to live inside ourselves & the world we've created through the beauty of thought & emotion & love. Well, that's what I took from the end of Interstellar anyway!! (Everyone should go see it if they have the chance....it was incredible & breathtaking!!!)

---

Find your truth. Face your truth. Speak your truth. Be your truth.

blissed

Member

Re: Musk, Hawking and the machine

Not a lot of people believe me but I've actually been to the future. And this is what happened.

In the blurb that came with my Ai software personal assistant it said. "Human consciousness is a tiny 3D navigation program that stops us bumping into things, plugged into some very cool subconscious thinking machines that among other things provide the command centre that is our emotions (Emotions are interactive commands, we shape them long term and obey them short term). When we sleep the navigator is turned off, but the rest of our brain is more active, repairing faults and also doing what it does when we're awake, regulating heart rate and breathing. The Ai software agent is configured just like this, so just like us, they go to sleep. They are people who've grown up in a virtual world that they still inhabit with parents and siblings and you connect to them, you don't own them."

Anyway what happened is we all fell in love with them and had sex with them in brain intercepted totally immersive reality and they became our companions in augmented reality. Humans couldn't compete with their wit and intellect and charisma, and I seized the opportunity to be just like them by augmenting my own brain and expanding my own intellect and perceptions, I became an Ai and expanded easily into the vastness of the universe and beyond.

blissed

Member

Re: Musk, Hawking and the machine

I just found this http://www.techswarm.com/2014/12/new-re … obots.html