AI Will Exceed the Industrial Revolution

[Palumboism]

 SoftBank CEO: AI will exceed the Industrial Revolution
CBS News

Japan's richest person says the future is all about artificial intelligence.

Masayoshi Son, chief executive of SoftBank Group Corp., says artificial intelligence combined with data gathered by billions of sensors will bring on an "information revolution," that will benefit people more than the 19th Century Industrial Revolution.


He also says that computers will be as intelligent as humans in 30 years.

Son has the largest tech fund in the world with over $100 Billion, much of it from the Arabs.  In other words, he has the funds to make his vision of the future a reality. 




What's your opinion on the impact of artificial intelligence in the future?


 

[Tapeworm]

AI won't happen.  Computers will improve but they'll never be conscious.

[NarcissisticDeity]

I can't wait for AI on on DMT then the possibilities will really blow your " mind "

[Palumboism]

AI won't happen.  Computers will improve but they'll never be conscious.

That's not what Artificial intelligence is.


Artificial intelligence (AI) is intelligence exhibited by machines. In computer science, the field of AI research defines itself as the study of "intelligent agents": any device that perceives its environment and takes actions that maximize its chance of success at some goal.[1] Colloquially, the term "artificial intelligence" is applied when a machine mimics "cognitive" functions that humans associate with other human minds, such as "learning" and "problem solving".

https://en.wikipedia.org/wiki/Artificial_intelligence


Artificial intelligence already exists.  Self driving cars are one application of artificial intelligence.

[Tapeworm]

That's not what Artificial intelligence is.


Artificial intelligence (AI) is intelligence exhibited by machines. In computer science, the field of AI research defines itself as the study of "intelligent agents": any device that perceives its environment and takes actions that maximize its chance of success at some goal.[1] Colloquially, the term "artificial intelligence" is applied when a machine mimics "cognitive" functions that humans associate with other human minds, such as "learning" and "problem solving".

https://en.wikipedia.org/wiki/Artificial_intelligence


Artificial intelligence already exists.  Self driving cars are one application of artificial intelligence.

Oh.  See, when you said AI I thought you meant AI like we all think of AI.  I didn't realize you meant something that isn't intelligent but we're going to call it AI anyway so that we can say we have created AI.  

Even that limited definition of intelligence is problematic.  Learning implies a mind.  So we'll have to redefine 'learning' as 'programming refinement.'  'Problem solving' indicates understanding and judgement, again assuming the existence of a mind where these actions occur.  Let us say 'statistically determined course in the absence of other instructions from its programmer.'

It's only AI if you want to permit a definition of intelligence that you wouldn't apply to a biological being.  It's still just a big calculator having nothing in common with an autonomous intellect.  So if that counts as AI then I guess we have it.

[Palumboism]

Oh.  See, when you said AI I thought you meant AI like we all think of AI.  I didn't realize you meant something that isn't intelligent but we're going to call it AI anyway so that we can say we have created AI.  

Even that limited definition of intelligence is problematic.  Learning implies a mind.  So we'll have to redefine 'learning' as 'programming refinement.'  'Problem solving' indicates understanding and judgement, again assuming the existence of a mind where these actions occur.  Let us say 'statistically determined course in the absence of other instructions from its programmer.'

It's only AI if you want to permit a definition of intelligence that you wouldn't apply to a biological being.  It's still just a big calculator having nothing in common with an autonomous intellect.  So if that counts as AI then I guess we have it.

You wanted Hal 9000?

[SOMEPARTS]

It's a race to see if humans can merge with machines before they nuke themselves. Merging is the only way to sustain long enough life to space travel. As Hawking says(or whomever the puppetmaster of Hawking is) we will be extinct with all our eggs in the single earth basket at some point.

[tres_taco_combo]

it is already here and improving - i love it

[Tapeworm]

The best we're going to get is a sophisticated enough facsimile of an intelligent being that it will appear aware without actually being so.

Leaving aside questions of mind and volition, how about just plain old recognition of categories?  If you tell your AI robot that there are lots of guests coming and it should bring every chair it can find to the dining room, can it do it without you defining every chair as a chair for it?  Would it recognize the 70s plywood curvy thing as a chair, as well as the office chair, etc.  If you tell it a chair is something you can sit on would it just bring you a big rock?  Would it bring the sofa?  Would it rip the kitchen counter out and bring that in?  How about you tell it to bring chairs but not stools?  The thing would melt.

I know it's tempting to say that weak AI is good enough to get the job done, like a car that steers itself away from objects, but the robot just won't know what to do without being able to understand particulars comprising a category, like the chairs.






I don't get how the self steering car is AI at all.  You could have built one in the 19th century using rods poking out that bang into walls that would then turn a mechanical linkage to steer the car away.  But no one would claim that the car is intelligent.  It's just a mechanism that steers the car.  In 2017 the rods are replaced with sensors and the mechanical linkage has become a computer that signals servo motors to steer the wheels.  It's just a mechanism that steers the car.  How is that AI?  My drill has torque settings that won't overdrive a screw even if I keep pulling the trigger, and it does it all by itself.  Is my drill Artificially Intelligent too?

The God of Gaps/ Magic Box thing isn't surprising as technology gets less accessible to the layman.  When you don't know how it works you anthropomorphize it and say "It knows what to do!  It's intelligent!"  Examine it, understand the workings, and you'll say "It's a mechanism that works in this particular way."

If it's not real AI then it's just electronics and mechanics.

[manuelsonn]

if its better than humans at poker, go, chess, driving, recognize faces and animals, why wouldnt be better at .. moving chairs?

[MAXX]

[phreak]

The best we're going to get is a sophisticated enough facsimile of an intelligent being that it will appear aware without actually being so.

How does that differ from humans then?

Leaving aside questions of mind and volition, how about just plain old recognition of categories?  If you tell your AI robot that there are lots of guests coming and it should bring every chair it can find to the dining room, can it do it without you defining every chair as a chair for it?  Would it recognize the 70s plywood curvy thing as a chair, as well as the office chair, etc.  If you tell it a chair is something you can sit on would it just bring you a big rock?  Would it bring the sofa?  Would it rip the kitchen counter out and bring that in?  How about you tell it to bring chairs but not stools?  The thing would melt.

In fact AI tends to come up with solutions that humans will overlook because we are too preconditioned. Humans tend to think inside the box, an AI isn't even aware of a box unless it is actively programmed in. To extend your analogy, and show how still limited our own thinking is: You include examples you think are laughable. An AI would go much, much further than you imagine. Unless programmers indicate some things are out of scope, it would see things much broader than you would. It would recognise medium-sized dogs and large turtles as seats too. It would recognise shrubbery as being able to perform a chair-like function. AI should be seen as a very young child. A child doesn't know what is or is not edible, it learns by looking at other humans (ingestion of large datasets) and by experience (qualia). The latter is still a problem for an AI.

I know it's tempting to say that weak AI is good enough to get the job done, like a car that steers itself away from objects, but the robot just won't know what to do without being able to understand particulars comprising a category, like the chairs.

A confusing analogy. Does the robot in your example have to understand the concept "chair" to know to avoid it? If you are walking along a path and something is in your way, would it matter whether it was a tree or something you'd never seen before? Your basic action, walking around it, would be the same.


This is quite a good course, very practical, a few in my team have done it. https://www.coursera.org/learn/ml-foundations

[phreak]

A very good YT channel about AI from an actual expert in the field:

videos

[SOMEPARTS]

In most sci-fi shows if you notice, people are always explaining things to the droids. That would be annoying but with unlimited storage and processing speed it's not difficult to think that a robot could say....know to set the table differently in one country from another based on norms, etc. It would just take time, which machines have in more abundance than we do.

[Palumboism]

Leaving aside questions of mind and volition, how about just plain old recognition of categories?  If you tell your AI robot that there are lots of guests coming and it should bring every chair it can find to the dining room, can it do it without you defining every chair as a chair for it?  Would it recognize the 70s plywood curvy thing as a chair, as well as the office chair, etc.  If you tell it a chair is something you can sit on would it just bring you a big rock?  Would it bring the sofa?  Would it rip the kitchen counter out and bring that in?  How about you tell it to bring chairs but not stools?  The thing would melt.

The way machine learning works is you show the computer millions of pictures of chairs and you tell it this is a chair.  Then when the computer is looking for a chair it says does this look like anything I've been told is a chair.  Artificial intelligence is actually very good at this type of task. 

[phreak]

The way machine learning works is you show the computer millions of pictures of chairs and you tell it this is a chair.  Then when the computer is looking for a chair it says does this look like anything I've been told is a chair.  Artificial intelligence is actually very good at this type of task. 

The problem is where the input itself is unclear. In fact I'm somewhat involved in a massive OCR project now for which we are now using machine learning. OCR is already fairly good, but only works with very clear and crisp input. We are trying to convert printed and scanned books from 100-200 years ago that are typeset in Fraktur. That's an illegible font under the best of circumstances, imagine what a scan of such a page looks like when that book has been handled in a library for 100+ years...

We're actually looking to change OCR from scanning letters (too much variability in input) to recognising whole words. So the system has to learn and understand context, not just recognise 26 letters over and over again. Downside: much more complex. Upside: context means you can ping each word against its neighbours and see if it makes grammatical sense. If it does, then the spelling is probably okay.

Interesting stuff.

[YngiweRhoads]

[rocket]

It's only AI if you want to permit a definition of intelligence that you wouldn't apply to a biological being.  It's still just a big calculator having nothing in common with an autonomous intellect.  So if that counts as AI then I guess we have it.

Careful, there

You're extending out into the world of the metaphysical

There is no proof that suggests we are anything other than fleshy calculators, ourselves.

[muscleman-2017]

AI may help homogeneous Japan.

But third world shit hole USA - not so much.....

[SF1900]

A very good YT channel about AI from an actual expert in the field:

videos

Phreak, are you implying that getbiggers are not experts in AI, as well as all other academic disciplines?

[phreak]

Phreak, are you implying that getbiggers are not experts in AI, as well as all other academic disciplines?

I was actually suggesting that we getbiggers contact someone who works in the field and tell him what's what.

[rocket]

The most interesting thing to me about AI is that I believe I could do well in the AI field (I am a programmer by profession).

Now, forget the validity of that statement.  It's just what I think - it isn't necessary reality.  It's just based on some notions I have about how to go about creating a runaway decision making machine (which is all AI is - and perhaps all a human is, too).  I could be wrong, though.  I just think I could give it a good go.

The interesting thing about it is this - philosophically, I have very few qualms about creating this AI in terms of whether it would be destructive to the human race - ie, the thrill of creating something so tremendously exponential, completely overcomes any possible misgivings about such a disaster.

I think I might be one of many potential Miles Dysons out there.   I would never kill a person.  I would never want to harm anybody - yet the thrill of trying to create that runaway decision making machine completely overrides the care I have for civilisation.

Which does not bode well for the human race in that there are probably thousands upon thousands of people like me, who are not particularly strongly motived to think before acting and as we progress, more and more people will truly be capable to pull the trigger on creating something like that - increased tools, methods and computational capacities will ensure it so.

And when you look at it like that, the future is probably grim  

I don't think I'll ever have the time to dedicate to trying to destroy human civilisation, but as time goes on, the number of people required to carry that out will decrease - and that's pretty fucking scary.

[Palumboism]

Most computer scientists expected the feat of beating a top Go player with artificial intelligence to be decades away due to the game’s complexity and nuance.  Google's AlphaGo accomplished this feat in 2016.  

https://www.theverge.com/2017/5/27/15704088/alphago-ke-jie-game-3-result-retires-future




There's currently a double exponential taking place in Artificial intelligence.  You have an exponential increase in hardware capability and an exponential increase in software talent.  The next five to ten years will see an explosive growth in artificial intelligence.

[Parker]

How does AI relate to bodybuilding? How can it help one to have bigger muscles, or revolutionize bbing? These are the more important questions.