The Future of Artificial Intelligence

That image is actually of "No.1"... he was a dumb robot.

A picture of the proper J5 would be more apt. :D

Johnny 5 still alive!

I know, thats a tag but "Don't they fear from it??". I mean everybody see the "Terminator series" or the "I Robot". What's the bond, that if we give the controll of the weaponary (that "we" created to kill each other) to a superintelligent computer ('who' makes killing more effective), that won't kill us??

Killing humanity is the most logical choise in favour of Earth. (And a comuter is only logic. Nothing more. No intuitions, no hope, no fear, no mercy...)

Hell!! The biggest danger to humanity is humanity itself...

(Sorry for my englis, and for my demagogy, but somebody needs to think about that.)

What a fascinating article, and brilliantly written to boot!

No mention of IMB's Deep Blue and Watson? Nor of ELIZA, ALICE etc or the Turing Test? Really?

What's there's well and informatively written, but there's so much more to cover!

An interesting read, but as mentioned above, it is a huge topic and there is so much more to write about! I hope you get round to writing some more stuff like this.

Also here is my universities version of the lab robot mentioned,
http://www.livescience.com/technology/090402-robot-scientist.html
video: http://www.youtube.com/watch?v=IY1sPV9e9H0

He built a 30-quad-core CPU rig, just to gambling, I think this is only a waste of energy. You know, calculating billions of "if...then..." states that 99% never actually happen

WOULD YOU LIKE TO BE A PEPPER TOO?

joking aside Ive always loved the concept of AI, even though the speilberg film back in 01 was utter tripe.

Enjoyable article but,"I need more input!"

Agree with TESCO-Zsömle:

Anybody seen Eagle Eye? :D

Very interesting read... the first 2 pages at least, didn't have time to read the rest yet. ;)

No Wintermute and Neuromancer?! For Shame! :p

Great article, I enjoyed that.
Like HugoB says, though, there's so much more to the topic that could be mentioned.
A follow-up article would be very welcome. :)

Great article. Thank you :)

NO-VA!

I wonder if the real future, will involve the embedding of computer technology into the human brain, allowing us to combine the huge data processing power of the CPU, with our capacity for abstract thought. Now that would be a potent combination!

Originally Posted by HugoB
No mention of IMB's Deep Blue and Watson? Nor of ELIZA, ALICE etc or the Turing Test? Really?

What's there's well and informatively written, but there's so much more to cover!

Exactly!

Like how about some actual data, for instance:

The highest current estimate for the computational power of the brain is 10 to the 16th computations per second. According to Moores law we'll have this for $1,000 in 2020, and for $1 in 2030...and storage - where a 1 neuron is equivalent to 1 bit - circa 300Tb.

Now, that's just hardware...it's our software that gives us intelligence. Given that it's only recently that computational power reached a critical point to allow biology to become a digital science (on par with chemistry and physics) and that it's increasing exponentially I see no reason why our understanding of biological processes shouldn't be sufficient to create intelligent life relatively soon.

Originally Posted by VipersGratitude
Exactly!

Like how about some actual data, for instance:

The highest current estimate for the computational power of the brain is 10 to the 16th computations per second. According to Moores law we'll have this for $1,000 in 2020, and for $1 in 2030...and storage - where a 1 neuron is equivalent to 1 bit - circa 300Tb.

Now, that's just hardware...it's our software that gives us intelligence. Given that it's only recently that computational power reached a critical point to allow biology to become a digital science (on par with chemistry and physics) and that it's increasing exponentially I see no reason why our understanding of biological processes shouldn't be sufficient to create intelligent life relatively soon.

It's interesting for the future though.

It's just a matter of time before computing power is enough to 'emulate' a human brain (just need to work out the software and how to download the data)

Once that happens that opens the floodgates to all kinds of interesting possibilities

(okay okay, i read too much science fiction, i can see it happening though, only 'when' is the question)

Just imagine if you could quicksave/quickload your brain, restore from backups...

I LOVE SHORT CIRCUITS!!!!

Its what got me into Robotics as a kid, so awesome I have the original tapes still and the DVDs ^__^

So the singularity is Ghost in the Shell -> Armitage -> Terminator -> Matrix? In that order?

Originally Posted by VipersGratitude
Now, that's just hardware...it's our software that gives us intelligence. Given that it's only recently that computational power reached a critical point to allow biology to become a digital science (on par with chemistry and physics) and that it's increasing exponentially I see no reason why our understanding of biological processes shouldn't be sufficient to create intelligent life relatively soon.

You're making some rash assumptions. So far as we know, the brain doesn't work like so simple a way as one neuron equating to one bit in a hard drive or one transistor in a CPU. It isn't a fixed form or function device.

There's a level of redundancy in the brain we're nowhere near emulating in silicon, too. I'd like to see a computer where you could rip out half of its components and the rest would re-wire to compensate. Although I suppose you could slightly analogise with a dual-CPU set-up - call the northbridge the corpus callosum and take each CPU as a hemisphere. Given appropriate design and software, it would - theoretically - be possible to disable one core without killing the PC. But as the brain is the CPU, hard drive and RAM all made out of the same components, we're still a long way from any soft of appropriate model. The idea that the brain evolved in (at least) RAID 1 does rather amuse me, though.

And that's just the hardware. The 'software' we run on is far less decipherable.

Until we understand exactly how the brain and mind do work we've no hope of emulating them. Biology is far from a digital science yet.

Originally Posted by HugoB
You're making some rash assumptions. So far as we know, the brain doesn't work like so simple a way as one neuron equating to one bit in a hard drive or one transistor in a CPU. It isn't a fixed form or function device.

Very true, I did break it down in to simplistic terms (with their inherent assuptions), but to give an entire explanation would be to outweigh the article itself.

Originally Posted by HugoB

I'd like to see a computer where you could rip out half of its components and the rest would re-wire to compensate.

See: The Internet :P

Alternatively do some research on fungible computing. I tried to google a more elaborate explanation of it, but the best I could find is in this video, 1min 20 secs in

Originally Posted by HugoB

And that's just the hardware. The 'software' we run on is far less decipherable.

Until we understand exactly how the brain and mind do work we've no hope of emulating them. Biology is far from a digital science yet.

You have to learn to walk before you can run. Barracelli's, and later Wolfram's research on cellular automata have shown emergent strategies to self-replication - intelligence inherent in the system. I doubt that the first AI will have human-level intelligence, but once we understand the underlying principles of intelligence our understanding of it will quickly evolve.

I'm always sceptical of this. Given we have such little true understanding of how our own brains work (see the entire field of Psychology), how are we going to make a computer do the same thing?

It's like trying to hum a song you've read the lyrics of, but haven't heard being played.

...quantum mechanical processes are considered to be at the basis of the reflective mind...enough computing power will [not] spontaneously produce consciousness

Just because it's quantum mechanics doesn't mean we can't simulate it - Even if the quantum effects are more than just an extra random factor (which we could simulate without using real quantum effects), the field of quantum computing is progressing rapidly. A quantum computer connected up like a brain would be significantly more powerful than a human brain.

Originally Posted by perplekks45
Anybody seen Eagle Eye? :D

I recently saw that! I didn't see any trailer for it so I thought it was going to be a bit of a flop for some reason. Now I'm glad I saw it because it turned out to be pretty good & well done for what it was. /off-topic :)

Ninja Edit: Great Article, that was quite an interesting read. Thnx.

Originally Posted by Bauul
I'm always sceptical of this. Given we have such little true understanding of how our own brains work (see the entire field of Psychology), how are we going to make a computer do the same thing?

It's like trying to hum a song you've read the lyrics of, but haven't heard being played.

More like trying to reconstruct a composer's entire ouvre, all the instruments of the entire orchestra that played it, and the electronics that recorded and broadcasted it, from having heard a few minutes of a single symphony on a radio.

Actually we know a reasonable amount of (neuro)psychology -- the gross outlines at least. People are a lot more predictable than we like to admit to ourselves. Ask Derren Brown. :)

I'll come back to this but I'm jotting on my tablet PC right now and that does not lend itself to convoluted posts. Handwriting recognition is scarily accurate but it is still a slow method of text input.