Login required to started new threads

Login required to post replies

Artificial intelligence will have an IQ of 10,000. We're saved from ourselves!
Quote | Reply
In the future the only jobs will be the ones robots can't perform:


Super artificial intelligence is coming, and sooner than you might expect.
That's according to SoftBank CEO Masayoshi Son. The Japanese billionaire spoke from the Future Investment Initiative in Riyadh, Saudi Arabia on Wednesday. In about 30 years, artificial intelligence will have an IQ of 10,000, Son says. By comparison, the average human IQ is 100 and genius is 200, according to Son.

Mensa, "the High IQ society," starts accepting members with an IQ score of 130.

The idea of machine learning becoming smarter than the human brain is often referred to as the "singularity." When exactly this will happen is oft-debated among the tech community.

https://www.cnbc.com/...-an-iq-of-10000.html

"The great pleasure in life is doing what people say you cannot do."
Quote Reply
Re: Artificial intelligence will have an IQ of 10,000. We're saved from ourselves! [jkca1] [ In reply to ]
Quote | Reply
Machine learning is radically different than AI. ignorant people make false prognostications.

Jim
"In dog beers, I've only had one"
http://www.shakercolonial.com/
Creating custom made furnishing to your requirements
Quote Reply
Re: Artificial intelligence will have an IQ of 10,000. We're saved from ourselves! [jkca1] [ In reply to ]
Quote | Reply
He's wrong.

AI is still awaiting its theory-of-relativity-grade breakthrough.

It still suffers from the problem that small increases in AI "intelligence" require exponential increases in software and hardware complexity.

So AI is good at very narrowly defined problems. Like chess. Or certain kinds of language processing.

Horrible at anything like general human intelligence. Take self-driving cars. Driving is that difficult for us. A very dumb person can drive a car. But we won't see a fully self-driving car for at least 5-10 years.

The AI community had a bunch of breakthroughs in the 60's and 70's. That's when all the main research on symbolic reasoning, etc, was done. There were huge expectations back then. But it kind of fizzled out. There was another surge of expectation in the 80's when "neural networks" were hot. Those didn't really pan out as "real AI" either (useful for many things, but they're definitely not a brain-surrogate).

Still waiting for the next hot thing. The latest hotness is "deep learning" which uses is an extension of neural networks that often takes advantage of newer processor architectures (e.g. GPU) to make better machine learning systems. But this is evolutionary, not revolutionary stuff. (in my opinion). E.g. it'll make very useful things (self-driving cars, really good voice and image recognition). But nothing at all like a 100 IQ human brain.
Quote Reply
Re: Artificial intelligence will have an IQ of 10,000. We're saved from ourselves! [trail] [ In reply to ]
Quote | Reply
trail wrote:
He's wrong.

AI is still awaiting its theory-of-relativity-grade breakthrough.

It still suffers from the problem that small increases in AI "intelligence" require exponential increases in software and hardware complexity.

So AI is good at very narrowly defined problems. Like chess. Or certain kinds of language processing.

Horrible at anything like general human intelligence. Take self-driving cars. Driving is that difficult for us. A very dumb person can drive a car. But we won't see a fully self-driving car for at least 5-10 years.

The AI community had a bunch of breakthroughs in the 60's and 70's. That's when all the main research on symbolic reasoning, etc, was done. There were huge expectations back then. But it kind of fizzled out. There was another surge of expectation in the 80's when "neural networks" were hot. Those didn't really pan out as "real AI" either (useful for many things, but they're definitely not a brain-surrogate).

Still waiting for the next hot thing. The latest hotness is "deep learning" which uses is an extension of neural networks that often takes advantage of newer processor architectures (e.g. GPU) to make better machine learning systems. But this is evolutionary, not revolutionary stuff. (in my opinion). E.g. it'll make very useful things (self-driving cars, really good voice and image recognition). But nothing at all like a 100 IQ human brain.

At some point there will be breakthroughs we can't imagine yet. If they come next month / next year the realization of the breakthroughs may not hit the general populace for 5+ years because they are not cost efficient to support. But if they involve an industry like investing or banking they may hit the street years sooner because the money needed to implement the solutions is far less then the potential profits.

Self-driving cars will one day arrive. The concept, once it becomes reality, will change the auto industry and insurance industry in ways we can barely imagine today in our self-centric gotta have a car world.

"The great pleasure in life is doing what people say you cannot do."
Quote Reply
Re: Artificial intelligence will have an IQ of 10,000. We're saved from ourselves! [jkca1] [ In reply to ]
Quote | Reply
I really hope that's the case. I'm still struggling to figure out how we get past the phase where there are both humans and self-driving cars sharing the roads. Seems to me that the humans will simply game the computers to the extent that in some cases they'll be practically immobilised. E.g. in heavy urban traffic where everybody is fighting for every inch of space and there are pedestrians, cyclists and motorbikes weaving through the traffic as well, I don't see how a self-driving car can cope while also having safety parameters that will be acceptable. I can absolutely see how they'll work fine in free-moving freeways (they're pretty much there already), or even in fairly free-moving suburban areas, just not in really heavy traffic.

There is a junction near me (right hand turn joining a road with 2 lanes going each way) where at busy times it's practically impossible to get out without being a bit of a dick. E.g. inching forward until you're sticking out enough that somebody takes pity on you and moves over into the far lane to let you in, or just gunning it into a gap that's not really big enough, and then waving a hand to apologise to the driver behind. There's also an art to it - if I'm going to gun it into a gap, I'll be looking at the driver who I'm about to piss off, and I'll be making a judgement call on a whole bunch of factors. Some a computer could deal with - is he or she cruising along, or are they accelerating to try and close that small gap down to nothing? Some are much harder than that - is he or she paying attention to the road, is there anything to indicate that the driver is likely to be a complete asshole who tailgates me for the next 5 miles or worse, and even doing calcs on how expensive and heavy their car is compared to mine if it turns out they're checking their email and they crash into me!

In a world of every car being self-driving these issues all go away (apart from pedestrian and cyclist interactions with cars), since you can establish standard protocols for filtering, assigning priority at junctions, etc so that it all flows smoothly. I'm just struggling to see how the transitional phase works. Maybe I'm overthinking it - we'll start out with cars with both a self-driving and human option, with the human taking over when things get too busy. Once cars with a self-driving option hit critical mass and the majority of miles are self-driving, and the benefits are starting to be seen, then we'll move to banning humans from driving completely (that's going to be an interesting debate/fight!). Assuming it takes another 5+ years for cars to have fully fledged self-driving capability, then at least another 10 years before they take over on the roads (probably longer, some people will fight this every step of the way), then I'm guessing it may be 20 years or more until we have cars that are totally designed to be self-driving with no human controls. Which is a shame as that's where the real benefits come in for me - designing a car entirely around passengers, and not drivers, will be awesome.
Quote Reply
Re: Artificial intelligence will have an IQ of 10,000. We're saved from ourselves! [jkca1] [ In reply to ]
Quote | Reply
What only 10 points higher than our leader?
Quote Reply
Re: Artificial intelligence will have an IQ of 10,000. We're saved from ourselves! [jkca1] [ In reply to ]
Quote | Reply
Quote:
artificial intelligence will have an IQ of 10,000

So about the same as what MSNBC tells us Obama’s is?

Civilize the mind, but make savage the body.

- Chinese proverb
Quote Reply
Re: Artificial intelligence will have an IQ of 10,000. We're saved from ourselves! [jkca1] [ In reply to ]
Quote | Reply
it's kind of a silly idea, in a way. IQ being a literal quotient of mental age over real age, a machine has neither, so it doesn't really have an IQ.

i think it's probably more useful to say that a machine can solve this equation X times faster than the average human or has a vocabulary X words larger. but to me to say that a computer has an IQ of 10 000 is a bit meaningless - what's the IQ of the encyclopedia britannica? what's the IQ of a slide rule? etc.

____________________________________
https://lshtm.academia.edu/MikeCallaghan

http://howtobeswiss.blogspot.ch/
Quote Reply
Re: Artificial intelligence will have an IQ of 10,000. We're saved from ourselves! [trail] [ In reply to ]
Quote | Reply
trail wrote:
He's wrong.


AI is still awaiting its theory-of-relativity-grade breakthrough.

It still suffers from the problem that small increases in AI "intelligence" require exponential increases in software and hardware complexity.

So AI is good at very narrowly defined problems. Like chess. Or certain kinds of language processing.

Horrible at anything like general human intelligence. Take self-driving cars. Driving is that difficult for us. A very dumb person can drive a car. But we won't see a fully self-driving car for at least 5-10 years.

The AI community had a bunch of breakthroughs in the 60's and 70's. That's when all the main research on symbolic reasoning, etc, was done. There were huge expectations back then. But it kind of fizzled out. There was another surge of expectation in the 80's when "neural networks" were hot. Those didn't really pan out as "real AI" either (useful for many things, but they're definitely not a brain-surrogate).

Still waiting for the next hot thing. The latest hotness is "deep learning" which uses is an extension of neural networks that often takes advantage of newer processor architectures (e.g. GPU) to make better machine learning systems. But this is evolutionary, not revolutionary stuff. (in my opinion). E.g. it'll make very useful things (self-driving cars, really good voice and image recognition). But nothing at all like a 100 IQ human brain.


I agree. This Softbank guy does a lot of wishful thinking, and is hyping things up to draw investors into his grand delusion.

Computers currently can only do exactly what humans program them to do. Even the biggest breakthroughs in pattern recognition today aren't even as good as the dumbest animals. The software required for computers to become intelligent enough to "think" on their own will be mind-boggling complex. This will require a massive increase in processing power. The problem with that is increases in CPU performance (aka Moore's Law) is starting to slow down. There's already talk about 2nm process being the practical limit. That isn't really a huge gain over the 10 and 12nm that's in use today, and there isn't any feasible breakthrough technology on the horizon to replace it.

Sure, large data centers can pool thousands of processors together and perform complex tasks, but not in real time, and that's not going to scale down to "IOT" devices. Some of the best "AI" we have can barely keep a car on the road without bumping into things. Despite years and years of working on it, the improvements are only incremental. A puny little bird brain is still better at navigation. Software is still buggy and inadequate. The best facial recognition barely works. The best voice recognition barely works. The most cutting edge devices out today are still rife with bugs and performance issues. Those devices still depend on us to take many cumbersome steps tell them what we need/what to do and the results are still garbage a lot of the time. There is nothing revolutionary out there.

Robots will only be able do simple tasks that are narrowly defined and repetitive for many years to come. All this talk of robots taking over the world soon is complete bullshit.

I'll just leave this right here:

https://www.extremetech.com/...kes-82944-processors

"It took 40 minutes with the combined muscle of 82,944 processors in K computer to get just 1 second of biological brain processing time"
Last edited by: RZ: Oct 26, 17 4:43
Quote Reply
Re: Artificial intelligence will have an IQ of 10,000. We're saved from ourselves! [cartsman] [ In reply to ]
Quote | Reply
cartsman wrote:
I really hope that's the case. I'm still struggling to figure out how we get past the phase where there are both humans and self-driving cars sharing the roads. Seems to me that the humans will simply game the computers to the extent that in some cases they'll be practically immobilised. E.g. in heavy urban traffic where everybody is fighting for every inch of space and there are pedestrians, cyclists and motorbikes weaving through the traffic as well, I don't see how a self-driving car can cope while also having safety parameters that will be acceptable. I can absolutely see how they'll work fine in free-moving freeways (they're pretty much there already), or even in fairly free-moving suburban areas, just not in really heavy traffic.

There is a junction near me (right hand turn joining a road with 2 lanes going each way) where at busy times it's practically impossible to get out without being a bit of a dick. E.g. inching forward until you're sticking out enough that somebody takes pity on you and moves over into the far lane to let you in, or just gunning it into a gap that's not really big enough, and then waving a hand to apologise to the driver behind. There's also an art to it - if I'm going to gun it into a gap, I'll be looking at the driver who I'm about to piss off, and I'll be making a judgement call on a whole bunch of factors. Some a computer could deal with - is he or she cruising along, or are they accelerating to try and close that small gap down to nothing? Some are much harder than that - is he or she paying attention to the road, is there anything to indicate that the driver is likely to be a complete asshole who tailgates me for the next 5 miles or worse, and even doing calcs on how expensive and heavy their car is compared to mine if it turns out they're checking their email and they crash into me!

In a world of every car being self-driving these issues all go away (apart from pedestrian and cyclist interactions with cars), since you can establish standard protocols for filtering, assigning priority at junctions, etc so that it all flows smoothly. I'm just struggling to see how the transitional phase works. Maybe I'm overthinking it - we'll start out with cars with both a self-driving and human option, with the human taking over when things get too busy. Once cars with a self-driving option hit critical mass and the majority of miles are self-driving, and the benefits are starting to be seen, then we'll move to banning humans from driving completely (that's going to be an interesting debate/fight!). Assuming it takes another 5+ years for cars to have fully fledged self-driving capability, then at least another 10 years before they take over on the roads (probably longer, some people will fight this every step of the way), then I'm guessing it may be 20 years or more until we have cars that are totally designed to be self-driving with no human controls. Which is a shame as that's where the real benefits come in for me - designing a car entirely around passengers, and not drivers, will be awesome.

I remember reading a story about the early days of the auto. There was a town with two cars and of course they collided...

The transition won't be seamless but think about all the drunks that kill and injure people, then add all the other accidents to that list. I believe that self-driving cars will be safer because they will be programmed not to exceed the speed limit, to drive slower in bad weather and maybe if shut off all together if the going gets to dangerous in the eyes of your computer. They will also have cameras on them that record everything around them, meaning should an accident occur the video should help law enforcement and juries decide who is really at fault. That could save millions.

I don't know when the reality of self-driving cars will happen. I do know that if they were safe I would want one today.

"The great pleasure in life is doing what people say you cannot do."
Quote Reply
Re: Artificial intelligence will have an IQ of 10,000. We're saved from ourselves! [jriosa] [ In reply to ]
Quote | Reply
jriosa wrote:
Machine learning is radically different than AI. ignorant people make false prognostications.


Banks and hedge funds that rely on artificial intelligence threaten to inject risks into the financial system that could exacerbate a future crisis, according to global regulators.

"The financial industry’s rush to adopt AI raises the potential that firms will become overly dependent on technologies that herd them toward the same view of risks and could “amplify financial shocks,” according to a study published on Wednesday by the Financial Stability Board, a panel of regulators that includes the U.S. Federal Reserveand European Central Bank.

“AI and machine learning applications show substantial promise if their specific risks are properly managed,” the FSB said in a report that called for additional monitoring and testing of robotic technologies designed to lessen human involvement. “Taken as a group, universal banks’ vulnerability to systemic shocks may grow if they increasingly depend on similar algorithms or data streams.”

The FSB, headed by Bank of England Governor Mark Carney, said that many of the technologies are being designed and tested in a period of low volatility in financial markets, and, as a result, “may not suggest optimal actions in a significant economic downturn or in a financial crisis.”

"The great pleasure in life is doing what people say you cannot do."
Quote Reply
Re: Artificial intelligence will have an IQ of 10,000. We're saved from ourselves! [RZ] [ In reply to ]
Quote | Reply
RZ wrote:
Robots will only be able do simple tasks that are narrowly defined and repetitive for many years to come. All this talk of robots taking over the world soon is complete bullshit.

I agree, computers lack the adaptability of brains for things like movement. The simplest animal is light years ahead of the most advanced computers in being able to create dynamic, adaptable movement.
Quote Reply
Re: Artificial intelligence will have an IQ of 10,000. We're saved from ourselves! [jkca1] [ In reply to ]
Quote | Reply
As my dad would say when I was young. "Balogna sausage"
Quote Reply