[Tyler]

Does AI really threaten the future of the human race?
5 December 2014, 12:00 am

The end of the human race - that is what is in sight if we develop full artificial intelligence, according to Stephen Hawking in aninterview with the BBC. But how imminent is the danger and if it is remote, do we still need to worry about the implications of ever smarter machines? BBC NewsLink.

Source: AI in the News

To visit any links mentioned please view the original article, the link is at the top of this post.

[8pla.net]

Every race has a finish line.

[Art]

So are you sort of saying like, "Tag!...we're IT!?"

[squarebear]

I thought Hawking was supposed to be smart? I guess he should stick to his own field.
The football playing robots at this year's Loebner Prize give an indication of the likelihood of an AI singularity.

[Freddy]

It's a ball ! It's a ball ! It's a ball ! System overload ! It's....a....ball...thud !   

[Art]

HAHAHAHAHAHAHAHA!!!! Just watched the video!!! 

[Don Patrick]

Well, yes and no. Soccer robots are typically highschool student projects, so not the height of AI. At the same time, Asimo and DARPA's robots are hardly more agile despite 10 years of improvement. It'll be some time before scientists realise that legs have terrible weight distribution.
The soccer robots illustrate an important point: If you put AI in a robot, it'll fall over. If you put AI in a hospital system, it'll turn off life support. If you put AI in air traffic control, planes will crash. If you put AI in weapons systems, it'll launch the nuclear warheads. All except the latter have happened, and all were caused by malfunctions in dumber-than-human AI. The chances that half-intelligent AI will do something stupid are several falling skies higher than the chances that an evolving superintelligence would develop human-like territorial issues and kill the greatest available resource on the planet. It would be extremely illogical and counterproductive. This does not compute.

Stephen Hawking is a genious in the fields on which he has extensive knowledge, but he is doing no-one a favour by telling a Boogyman tale. Far better had he said "With great power comes great responsibility."

I think it's time I write another blog, and maybe program an actual super-AI virus just to illustrate a point. I'm getting tired of this doomsaying trend all year.

[Art]

Wow! Very well put Don!! My sentiments exactly!

In line with what you have written, if all of these AI systems failed, then the very Humans that wrote and programmed them, Failed, and Failed miserably! These AI have no conscious ability to discern all possible "what-if" scenarios or "should I do this" decisions. Why? Because obviously, they weren't programmed to allow for this in the first place. While some robots / AI can be said to "think", their thinking is usually very predictive and linear. Outside the box thinking is not their forté and it will likely be a long time before it is.

Yes, I too have great respect for Mr. Hawking but stirring the pot of doomsday is hardly scientific.

[8pla.net]

Some of these comments may demonstrate Stephen Hawking's point
(the limits of slow biological evolution).

[Art]

No, sorry but I would rather go with Ray Kurzweil's:

"Non-biological intelligence is growing exponentially. Biological intelligence isn't really growing at all or if it's growing, it's growing at such a slow rate that it's not noticable. Which is why non-biological intelligence ultimately will become dominant."

Yes, in time, I'm certain it will happen but not necessarily in OUR lifetime / generation. I'd like to be able to wait around long enough to be proven wrong, but I'm afraid that's not possible either! 

[Don Patrick]

Actually, neural nets for example have an exponential DEcline in their rate of improvement. Once they've seen a million examples of a pattern and reached a 50% accuracy or so, it then takes gazillions just to raise that accuracy by 1% further. This is a documented phenomenon, and it seems to indicate a ceiling to how far one can train a neural net.
So I would be skeptic about the statement that AI is growing exponentially. If it has, how odd then that most experts say that it's barely gotten off the ground since the 50's. Kurzweil may be thinking of computer speed growing exponentially, but the ceiling for that has also already been estimated on grounds of the minimum size of the core conductive material: one atom thick.

[Art]

I think he might be referring to the rate of improvement that non-biological (digital) intelligence has shown compared to human intelligence within the same time reference. Just 30 years ago, where were computers and computing? Practically in the dark ages. Where were people? Pretty much in the same place as we were then only 30 years older, more wrinkles, gray hair, some deceased, etc.

Since we really have no crystal ball, our speculation is...our speculation. Time will tell.

[Freddy]

Kurzweil is a futurist, sometimes I think it helps to sell his books or movies to go a little Sci-Fi 

I do think computing has come an extremely long way in a short time though. Just compare today's graphics to the state of the art in the 80s. Ray-tracing and physics simulations and all that CPU hungry stuff. And that's just one example.

[Don Patrick]

Ah, like that. Well, 60 years ago there were no computers so in comparison it's indeed gone rather fast. On the other hand if you look at car engines from start to now, major improvement happened within a century but kind of levelled out: Once we got the concept down all we're left with now is a bit of fine-tuning the fuel efficiency.

I am counting on time still being a factor: I have yet to see an evolutional AI that wasn't literally flollopping about on the floor. I don't think they'll stumble on the perfect evolutional model right away.

[8pla.net]

Well, 60 years ago there were no computers so in comparison it's indeed gone rather fast.

Mentioned, by no means intended as a correction, but solely for the ancient technology angle which is so interesting to discuss.  Computers are truly ancient devices.  For example, computers,  at the very least, predate Christianity.  The Antikythera Mechanism is an ancient computer.  They found the Antikythera Mechanism on a ship wreck. 

This suggests that ancient computers may have been built by ancient Greek computer engineers to supply entire fleets of ships.  Or even more, perhaps ancient computers were being shipped as technology products for trade.  This could make the ship an ancient floating computer store.

Ancient customers would line up, docked to the ship in row boats, for the latest ancient must have computer gadgets.  For formally educated computer scientists and self educated computer geeks alike... It is fascinating that computers at least existed on Earth before the time of Jesus Christ, historically speaking.  Who knows how much further back in time ancient computers existed.  Scientists are now saying the their estimation of the Neanderthal extinction was off by ten thousand years.  But that is for another topic.