When will human-level AGI be created?

[Kontinuum]

AGI = Artificial General Intelligence
I'd say it will be 2075-2100.

[PC2]

No clue. I think it is possible it could be hundreds of years away. On the other hand, I also think it is possible we may soon have the hardware and could be only one profound algorithm away, possibly reaching it in 10 years. It is hard to say when we don't know what breakthroughs will come.

For those interested, Elon Musk and others started a non-profit AI organization yesterday with $1 bil in funding. That is a huge. Elon is overly fearful though, so I'm not sure how much that would influence the organization.

https://www.openai.com/blog/introducing-openai/

[Glorious Leader]

No or at least it won't be understood as "human lvl". Human cosciousness, cognitive functions, the mind itself is not so easily reduced. Emergent properties exist and it's not that such functions can be replicated even assuming we had a profound understanding of the human mind.

[Very Tired]

2287. They'll be called Synths and everyone will freak out because they blend into normal human population seamlessly.

[Kraenen]

I have reports coming in right now saying 2051

[Cæli]

You need consciousness for that. Our species existed for about 200000 years, don't expect for us to recreate what nature created in less than 200000 years. 1000 years is absolutely nothing, the only great (and I mean great) discoveries before year 3000 will happen by accident, and odds of it happening are near to none.

[May90]

We still cannot even evaluate how complex such an AGI would be technologically. It is possible that our technological advancement in the nearest future will experience a sudden jump, and we will have built an AGI by 2030s or so - but my guess would be 2100s.

[PC2]

You need consciousness for that.

Maybe, but even if consciousness is uniquely carbon protein based, there is no reason to think organics couldn't be engineered towards super human AI. Technically anything we make can be considered "artificed", except for natural babies.

[Cæli]

Maybe, but even if consciousness is uniquely carbon protein based, there is no reason to think organics couldn't be engineered towards super human AI. Technically anything we make can be considered "artificed".

AGI needs to have cognitive abilities similar to humans and for that it needs to be conscious. Maybe the question is if we recreate an exact copy of someone, with the exact same matter arrangement (we're not even sure of what reality and matter is made of by the way), will this body be conscious of itself, or will it lack something ?

[GeneralWolfie]

Considering Nasa and Google just created a processor that is upwards of 100 million times the calculations p/sec than the average PC
and it is specially designed to handle algorithm testing, it possible it could be relatively right around the corner 2050 is not out of the ballpark by any means.

We are now dealing in the exponential tech curve with Quantum processors.

Source: http://www.telegraph.co.uk/technolog...n-your-PC.html

[Vash The Stampede]

I believe that true AI would arise once we know how the human brain works. Currently right now we know nothing how the human mind works but I believe that the Institute from Fallout 4 is a good example in how AI would evolve, except they left out sexbots which is going to be the driving force of the creation of these Androids.

Within 70 years
Gen 1 Sexbot - Pretty much a giant barbie doll with Siri or OK Google built into it. Might as well be a blow up doll with a voice box.
Gen 2 Sexbot - Able to walk and talk and interact with people better but still going to fall down and look like a rag doll doing it.
Gen 3 Sexbot - This is where these machine are so good that they walk to the store and even buy groceries as well as clean the house. Still not AI but enough that at a glance you couldn't tell.

Within 150 years
Gen 1 Android - This is basically a machine like from Aliens where it's a robot with fluid that is pumped throughout their body and has early AI like brain. Hard to tell it's not human.
Gen 2 Android - More advanced than the previous generation and has real AI. So much so that they have their own thoughts and may even fight for rights.

Within 200 years
Gen 1 Organic Android - Like in Fallout 4, people will use human DNA to construct these. They will have bones, and blood, and cells just like a human. In fact, they're probably human but with bits of machinery to keep them obedient. At some point both humans and these Androids will merge.

[Dundebuns]

Considering Nasa and Google just created a processor that is upwards of 100 million times the calculations p/sec than the average PC
and it is specially designed to handle algorithm testing, it possible it could be relatively right around the corner 2050 is not out of the ballpark by any means.

We are now dealing in the exponential tech curve with Quantum processors.

Source: http://www.telegraph.co.uk/technolog...n-your-PC.html

I believe it's only faster doing certain kinds of operations, and that it's the same speed as a standard machine doing standard things.

Don't forget that IBM invented Watson. It's certainly a step in the direction of AI

[Garnier Fructis]

I'm convinced that some computers are smarter than some people already.

[Deleted]

Well, considering that our phones today are more advanced than the best technology that landed us on the moon and that there's already some attempts towards artificial intelligence by reproducing what humans say and remembering the results, I'd say in about 2040-2060 we'll have an artificial intelligence smarter than most humans.
However, if we take the dumbest humans (not those that have little brainpower due to diseases, but the actual dumb people), the current artificial intelligence is already ahead.

[Lochton]

AGI = Artificial General Intelligence
I'd say it will be 2075-2100.

Depends on the human we calculate from as a standard example. If you understand my trail of thoughts.

[Kontinuum]

The difference between dumb humans and smart humans is much smaller than the gap between apes and (dumb) humans.

[anntheman]

I'm convinced that some computers are smarter than some people already.

I agree. Watson is probably smarter than the majority of people. One might even argue that perhaps a simple Bayesian, bag of words, etc. algorithm hooked up to the WWW is smarter than some people...

[Deathquoi]

I believe that true AI would arise once we know how the human brain works. Currently right now we know nothing how the human mind works but I believe that the Institute from Fallout 4 is a good example in how AI would evolve, except they left out sexbots which is going to be the driving force of the creation of these Androids.

Within 70 years
Gen 1 Sexbot - Pretty much a giant barbie doll with Siri or OK Google built into it. Might as well be a blow up doll with a voice box.
Gen 2 Sexbot - Able to walk and talk and interact with people better but still going to fall down and look like a rag doll doing it.
Gen 3 Sexbot - This is where these machine are so good that they walk to the store and even buy groceries as well as clean the house. Still not AI but enough that at a glance you couldn't tell.

Within 150 years
Gen 1 Android - This is basically a machine like from Aliens where it's a robot with fluid that is pumped throughout their body and has early AI like brain. Hard to tell it's not human.
Gen 2 Android - More advanced than the previous generation and has real AI. So much so that they have their own thoughts and may even fight for rights.

Within 200 years
Gen 1 Organic Android - Like in Fallout 4, people will use human DNA to construct these. They will have bones, and blood, and cells just like a human. In fact, they're probably human but with bits of machinery to keep them obedient. At some point both humans and these Androids will merge.

Good example of someone that has consumed far too much science fiction in their life.

Sure bud, real life will be just like Alien and Fallout.

[Independent voter]

Probably within 50 years, for sure a 100.

What would be cool is if they could take a snapshot of our brain and record it's state and then upload that data to a computer. We wouldn't need to understand consciousness and AI, we would just be recording a consciousness like taking a photo of something. When you take a photo, you don't need the talent to paint the scene, the camera does all the work.

[May90]

Probably within 50 years, for sure a 100.

What would be cool is if they could take a snapshot of our brain and record it's state and then upload that data to a computer. We wouldn't need to understand consciousness and AI, we would just be recording a consciousness like taking a photo of something. When you take a photo, you don't need the talent to paint the scene, the camera does all the work.

The camera might have to have an extremely high resolution though... One that might not necessarily be achievable technologically. We don't know everything about how the brain works yet, but the processes happening there occur on such extremely small scales that we might not even be able to "see" them experimentally, let alone copy.