creative AI

[Kafein]

Assuming a framework based on human minds, the issue is hardware.

Human brain design is not efficient, much like human ankle design. Based on that, an AI could be created that's more efficient, which suggests that the issue is software design rather hardware constraints.

If you have a human-level AI that can rewrite it's own code, it would be able to constantly improve itself. Since it may not have as many inefficiencies as regular human minds, it would be able to change itself fairly (it takes humans years to change their habits- an AI could do this in a second) quickly. Once it does this, it doesn't just become a particularly smart human- it's more like the difference between a dog and a human. Is a human going to listen to a dog, or do everything that a dog wants?

So you get an AI like this and you think 'O, let's just keep it isolated with no access to the outside world'. Think about how a human being is able to get lesser intelligences to do what it wants. Now, imagine something a degree smarter and how it might be able to persuade the humans it comes in contact with to give it external access or keep it running. What if it figures out that one of the researchers in the lab has a daughter that's dying of neuroblastoma, and offers a cure, if only it had access to more data? That doesn't even take advanced persuasive techniques into account. Stuff we haven't even mapped out yet.

That's one possibility. There's also the idea of a paperclip maximizer. Say you build an AI with the goal of making as many paperclips as possible (you could replace with 'farming as much nutrient dense food', or whatever) and find that AI constantly finds new ways to turn matter into paperclips. Where will it stop? Will it continue converting the entire solar system into paperclips?

The current expert estimates are between 20 and 250 years, where 20 is improbable.

Intelligence doesn't inherently require free will the way we understand "free will", which is already a flawed concept.


The paperclip problem is a real one but it essentially always has been. The first models of assisted driving in cars had no force feedback and the sensitivity wasn't adaptive, so essentially you could drive really fast and instantly turn your front wheels completely with only a tap on the wheel. That turned out to put too much pressure on the human. We already use drones in warfare which highlights the real problem of machines giving previously unattainable power to individual humans still using their tribal brains.

To create the technological singularity, ie. 'strong AI', science would first have to understand metaphysical terms such as 'mind' and 'thought', and we are a long, long way from that.

It's called the Turing test and we already have it. We are sure that humans have a "mind" and "thoughts" therefore if something behaves like a human then it must also have these things.

[Prinz_Karl]

Some people here have been reading/watching too much sci-fi. AI like that simply is not happening in the foreseeable future.

I agree. Especially AI that becomes self-controlling, even say self-productive and that will undermine human species will not happen if not someone really aims to construct such AI he can't controll himself.

What I mean is that human also didn't pop up self-thinking and independent, it was nature that formed our human brain and so it would need human (indirectly nature) to form self-thinking and independent AI. Only difference is that nature doesn't care if it's building something that's self-destructive.

[Algarn]

Some people here have been reading/watching too much sci-fi. AI like that simply is not happening in the foreseeable future.

What are you talking about? What development? We are not closer to true AI than we were 50 years ago. It's not a matter of just building more powerful computers.

It's not going to happen soon, afaik. But it will happen eventually. The "weak" AI era we're into is just the beginning, with machines executing simple tasks with algorithms more or less complex. I'm not an AI expert, so I can't tell how to build a better AI, but some will know in the future, doesn't matter if it's 50 years or 250 years after today.

[Christo]

The way I can even imagine having a sentient AI is.. well we would need to have a total understanding of the human brain and ALL of its functions, then copy and put those in a supercomputer.

Not gonna' happen for quite a long time.

Yeah totally not scientifically accurate and sci-fi explanation from me, whatever. 

[Xant]

It's not going to happen soon, afaik. But it will happen eventually. The "weak" AI era we're into is just the beginning, with machines executing simple tasks with algorithms more or less complex. I'm not an AI expert, so I can't tell how to build a better AI, but some will know in the future, doesn't matter if it's 50 years or 250 years after today.

How do you know that if you're not an AI expert?

[Algarn]

How do you know that if you're not an AI expert?

How can't one know that the current machines run on "basic" algorithms that make them look like they have a form of intelligence, and will be developed further, like everything nowadays ?  Being aware of that doesn't require much.

[[ptx]]

...

Do you understand what algorithms are? How they work? How software is built and functions?

[Algarn]

...

Do you understand what algorithms are? How they work? How software is built and functions?

Yes, a bit. I guess I didn't express myself well. An algorithm is a suit of commands, in some cases, these suits are complex, while on a simple calculator, they are much simplier. Even if I wasn't good at maths, I know a bit on them.

[Xant]

How can't one know that the current machines run on "basic" algorithms that make them look like they have a form of intelligence, and will be developed further, like everything nowadays ?  Being aware of that doesn't require much.

That has nothing to do with AI.

[The_Bloody_Nine]

I too think that something fundamental is missing to create real AI.

This leaked email or what though from dear Elon is interesting. From what I understood of the small text he claims that the progress of improving/learning of network AIs from companies like DeepMind is close to exponential and thus he's expecting some crucial point in the near future where 'something dangerous' could happen. But unless he explaines more in detail what he is exactly referring to I doesn't mean much for me. And I'm not sure if his worries are based on some knowledge we don't have or simply a bad feel in the gut. I mean, it makes only sense to expect the worst when actually working on AI rather than just taking for granted all will go well. Even if there are no concrete signs right now that, when or how things could get out of control.

[Beauchamp]

in 70 years of constant development we got a software that can distinguish dog's poop from tortoise and in the next 20 we can expect that some computer will be aware of itself?

https://www.youtube.com/watch?v=aboZctrHfK8

[Angantyr]

It's called the Turing test and we already have it. We are sure that humans have a "mind" and "thoughts" therefore if something behaves like a human then it must also have these things.

The Turing test has nothing to do with understanding these concepts, or recreating them. Which is what I was referring to. We don't even understand the relation or the boundaries between the physical and the non-physical, 'the mind-body problem', let alone complicated notions such as process of thought.

[Algarn]

That has nothing to do with AI.

But what is artificial intelligence then ? How would you describe it ? I describe it as a form of intelligence created by humans, and now, we can only create"AIs" which are barely as smart as mouses. Need a quote there, but it's approximately what we can do now, a really basic AI with limited "choices", while a real intelligence could not have these limitations, because it could reflect on itself and its environment, and therefore create "choices'' by itself.

Fuck english and french education on foreign languages on a side note.

[Golem]

Now, this is going into the deep(and shady) a little bit, but bear with me.
Imagine you were created, just now. All you are aware of is your being, your existence. The only thing you can do with this knowledge is to reflect on it. By reflecting on existence, you create the opposite within yourself. Non-existence, nothingness, space. Whatever you want to call it. Soon you realise that you are being engulfed by it. So you try to fill in the blank with something. Suddenly - life. An idea that was to save you but turns out to be just an illusion. Even though you supply it with heat, gamma rays and what not and it seems to prosper and fill in all the inked spots, the very nature of it is abyss-forming, for it is conscious - and creates blank space within itself. You made this from your image and just like you it succumbed to error.
What happens when life starts makin the universe to fit its image? Without your intervenience - a catastrophe!
But if you put your 'hands' in it, and lets assume that you did, life would start to create unconscious beings. Beings uncapable of creating blankness. An ongoing code without stop or loss of tempo. And once it has connected every planet every star and every galaxy you would hatch. Your eyes would be open or else manufactured at the borders of universe to predict and interact with other universes. Now get on EU1. A whole network of sentient mechanisms, trying to control each others development. Zoomout and you can see the whole sharade. Flickering lights dancing on the top of a bottomles mountain, with a quick explosion every now and then. Bang!


tl;dr the development of AI is controlled by the universe at large as is to an extent our own behaviour and strive

[Xant]

But what is artificial intelligence then ? How would you describe it ? I describe it as a form of intelligence created by humans, and now, we can only create"AIs" which are barely as smart as mouses. Need a quote there, but it's approximately what we can do now, a really basic AI with limited "choices", while a real intelligence could not have these limitations, because it could reflect on itself and its environment, and therefore create "choices'' by itself.

Fuck english and french education on foreign languages on a side note.

https://intelligence.org/2013/08/11/what-is-agi/