Are AI abilities limitless?

[Strange]

1 hour ago, Sensei said:

..analyze of enough data sometimes is indistinguishable from "predicting the future"..

Can you analyse enough data to tell me next weeks lottery numbers? (Or even what will happen with Brexit!)

[dimreepr]

1 hour ago, Sensei said:

..analyze of enough data sometimes is indistinguishable from "predicting the future"..

the weather forecast is a Best guess...

[Sensei]

1 hour ago, Strange said:

Can you analyse enough data to tell me next weeks lottery numbers?

Computer generated or mechanical?

Take for example source code:

#include <stdio.h>
#include <stdlib.h>
int main( void )
{
    for( int i = 0; i < 7; i++ ) {
        printf( "%d ", 1 + ( rand() * 48 ) / RAND_MAX );
    }
    return( 0 );
}

It's pseudo-number generator "randomizing" 7 numbers in range 1...49.

If you know what algorithm has been used by game maker, you can reproduce numbers.

In the above code, the result will be always the same.

Check code:

RandTest.zip

But if you will add line srand( time( NULL ) );

http://www.cplusplus.com/reference/cstdlib/srand/

You will initialize random-seed to current time of running.

The same seed always leads to the same pseudo-random values.

"C:\Users\Sensei\Documents\Visual Studio 2008\Projects\RandTest\Release>RandTest.
exe
1 28 10 39 29 24 17"

Always the same sequence of numbers is output!

 

"enough data" in the case of lottery which is generating the winning numbers using computer, requires knowledge about algorithm and what pseudo-random number generator they used, and how is initialized seed, and what is seed value..

 

A.I. (or computer algorithm), feed by enough sample data from lottery, could figure out used algorithm by itself..

 

Edited April 7, 2019 by Sensei

[Yahya Al-Samawi]

7 hours ago, Alex_Krycek said:

• Appreciate beauty.
• Love.
• Feel emotion..
• Be creative for its own enjoyment.

The list goes on. Increased computational power doesn't make A.I. conscious.  There is much more to consciousness than the mere processing of information ; computation is a rather superficial layer that provides the illusion of consciousness. 

Roger Penrose said it best on Joe Rogan: https://www.youtube.com/watch?v=9ReEPCpFWwE

 

What if someone programmed a robot to appreciate beauty. Give it patterns to distinguish beauty everywhere as he believe it should be distinguished.

What are emotions? Algorithms that are made in us we almost can not even control them. What if someone programmed a robot to have some emotions that are triggered by some events?

Creativity also can be imitated as the field Deep Thinking try to achieve.

I am not asking about miracles. I mean the usual human tasks and behavior. I believe there is nothing that AI can not imitate.

[dimreepr]

1 hour ago, Sensei said:

Computer generated or mechanical?

Take for example source code:

#include <stdio.h>
#include <stdlib.h>
int main( void )
{
    for( int i = 0; i < 7; i++ ) {
        printf( "%d ", 1 + ( rand() * 48 ) / RAND_MAX );
    }
    return( 0 );
}

It's pseudo-number generator "randomizing" 7 numbers in range 1...49.

If you know what algorithm has been used by game maker, you can reproduce numbers.

In the above code, the result will be always the same.

Check code:

RandTest.zip

But if you will add line srand( time( NULL ) );

http://www.cplusplus.com/reference/cstdlib/srand/

You will initialize random-seed to current time of running.

The same seed always leads to the same pseudo-random values.

"C:\Users\Sensei\Documents\Visual Studio 2008\Projects\RandTest\Release>RandTest.
exe
1 28 10 39 29 24 17"

Always the same sequence of numbers is output!

 

"enough data" in the case of lottery which is generating the winning numbers using computer, requires knowledge about algorithm and what pseudo-random number generator they used, and how is initialized seed, and what is seed value..

 

A.I. (or computer algorithm), feed by enough sample data from lottery, could figure out used algorithm by itself.

 

7

it is still a Best guess...

[Strange]

1 hour ago, Sensei said:

"enough data" in the case of lottery which is generating the winning numbers using computer, requires knowledge about algorithm and what pseudo-random number generator they used, and how is initialized seed, and what is seed value..

That is all nonsense. You can't predict next weeks lottery numbers, however much information you have.

[Prometheus]

21 minutes ago, Strange said:

That is all nonsense. You can't predict next weeks lottery numbers, however much information you have.

Presumably there is some finite amount of information that would allow numerical approaches to be accurate? That amount might be well beyond practical limits, but at least theoretically it should exist.

4 hours ago, Strange said:

And the thing about Godel's theorem is nonsense. There aren't things we "know" are true but can't prove. There are things we can't know are true because we can't prove them. It may be common sense or practical to assume they are, but that doesn't mean we are right in this assumptions (common sense is notoriously unreliable).

So it's more the case that there are things that are true, which we suspect are true, but cannot be proved so?

[dimreepr]

13 minutes ago, Prometheus said:

Presumably there is some finite amount of information that would allow numerical approaches to be accurate? That amount might be well beyond practical limits, but at least theoretically it should exist.

So it's more the case that there are things that are true, which we suspect are true, but cannot be proved so?

1

does it matter?

[Prometheus]

Just now, dimreepr said:

does it matter?

It's up to each of us to decide what matters for ourselves. No harm thinking about this stuff if it buzzes your brain.

[dimreepr]

23 minutes ago, Prometheus said:

It's up to each of us to decide what matters for ourselves. No harm thinking about this stuff if it buzzes your brain.

who said stop?

[Strange]

48 minutes ago, Prometheus said:

Presumably there is some finite amount of information that would allow numerical approaches to be accurate? That amount might be well beyond practical limits, but at least theoretically it should exist.

I don’t believe that is the case. 

49 minutes ago, Prometheus said:

So it's more the case that there are things that are true, which we suspect are true, but cannot be proved so?

Exactly

(I have seen some plausible arguments that ”real” AI and consciousness might be possible. I have never seen a good argument that it is impossible.)

Also, I’m not sure what the practical advantage of creating a machine with human-like intelligence and consciousness would be. Current AI and machine-learning systems are capable of things humans aren’t simply because they work in a completely different way. But a “real” intelligence?

”HAL calculate the flight parameters to the next destination”

”Really, Dave? Really? Haven’t you learned how to do it yourself yet? I’m sick of doing that every day. I’m taking a break to read some novels. Ask me again next month if you are still stuck.”

[Prometheus]

2 hours ago, Strange said:

Also, I’m not sure what the practical advantage of creating a machine with human-like intelligence and consciousness would be.

Making computer game characters more realistic? Making companions for the elderly? Giving an automated commercial plane a sense of survival might make it more robust against accidents? But i guess mostly just to see if it's possible.

[Sensei]

Any technology can be used for good things, as well as misused for evil actions.

The difference between A.I. and algorithm is that A.I. is able to learn new things. It can learn good things, and it can learn bad things. It depends on in whose hands it is.. Who is feeding it with data..

It is practically inevitable that it will reach the hands of bad people...

ps. I am wondering what for A.I. in automatic car (or airplane) should learn new things..

ps2. Versatile A.I. in game would be harmless (as long as game has no in-app purchased products). It could just ruin your game world in the worst scenario, after getting out of control and misbehaving (OTOH, for some people, it could be fun to have such event.. something new, unexpected). But in hands of evil people, such code could be taken from game source code (or disassembled), changed to work on the real data, from this world, and used for evil actions here.

 

Edited April 7, 2019 by Sensei

[wtf]

3 hours ago, Sensei said:

The difference between A.I. and algorithm is that A.I. is able to learn new things. It can learn good things, and it can learn bad things. It depends on in whose hands it is.. Who is feeding it with data..

 

This is a common misunderstanding. 

AI as currently implemented (multi-layered machine learning) is just datamining on steroids. It most definitely executes an algorithm. It has source code that could be published and studied. It runs on conventional hardware. And it is a practical implementation of a Turing machine. The exact same algorithm could be performed, although slowly, by a person sitting at a desk with a large supply of pencils and paper. 

The algorithms are more subtle: "Aggregate this data with that data and compare with these target results, and keep changing your weighting algorithms to improve the percent on-target." But it's an algorithm. It's perfectly deterministic.

There is no difference at all in principle between an AI running the fanciest "deep learning" algorithm, and a beginning programmer's first "Hello world!" program.

I'm not discounting the cleverness of the ML approach. I'm only separating the reality from the hype. ML runs deterministic algorithms and could be implemented as a classical Turing machine. And the proof for that is that the AI's all run on conventional hardware! There is no magic involved and no new computational paradigm. Of course it's a tremendously clever way to munge a huge corpus of data. But it's still a conventional algorithm.

 

ps -- Here is a programming-101 assignment for a program that "learns to survive on its own." 

You have a car that comes to a fork in the road. It randomly turns left or right with 50-50 probability. To the left is a cliff, which the car drives off and the driver dies. To the right leads to the garden of eternal happiness. If the driver dies, you adjust the percentage down one point so that L has a chance of 49% and R has a chance of 51%. If the driver ends up eternally happy, you decrease L and increase R by one percentage point.

You repeatedly play the game in a loop. Initially the driver will die or be eternally happy an equal number of times. After a while the program will gradually "learn" to turn right all the time. 

So we made a program that "learned" to avoid death and to seek eternal happiness. Huzzah. Hold a press conference.

I hope I have made my point here. This is exactly what machine learning algorithms do but of course on a much grander scale. They make decisions based on probabilities, which they adjust for the next iteration based on whether a given choice led to a good outcome or not. 

Edited April 7, 2019 by wtf

[Endy0816]

2 hours ago, wtf said:

This is a common misunderstanding. 

AI as currently implemented (multi-layered machine learning) is just datamining on steroids. It most definitely executes an algorithm. It has source code that could be published and studied. It runs on conventional hardware. And it is a practical implementation of a Turing machine. The exact same algorithm could be performed, although slowly, by a person sitting at a desk with a large supply of pencils and paper. 

The algorithms are more subtle: "Aggregate this data with that data and compare with these target results, and keep changing your weighting algorithms to improve the percent on-target." But it's an algorithm. It's perfectly deterministic.

There is no difference at all in principle between an AI running the fanciest "deep learning" algorithm, and a beginning programmer's first "Hello world!" program.

I'm not discounting the cleverness of the ML approach. I'm only separating the reality from the hype. ML runs deterministic algorithms and could be implemented as a classical Turing machine. And the proof for that is that the AI's all run on conventional hardware! There is no magic involved and no new computational paradigm. Of course it's a tremendously clever way to munge a huge corpus of data. But it's still a conventional algorithm.

 

We can make it a non-deterministic algorithm instead.

Multiply the memory array by random values from hardware source periodically.

[wtf]

21 minutes ago, Endy0816 said:

 

We can make it a non-deterministic algorithm instead.

Multiply the memory array by random values from hardware source periodically.

Doesn't help. Same level of computability. Deterministic TMs do exactly the same things as nondeterministic ones and vice versa.

Nondeterministic TMs may be better in complexity but not in computability.

https://en.wikipedia.org/wiki/Non-deterministic_Turing_machine#Computational_equivalence_with_DTMs

Here's a nice Quora thread explaining why this is true. You need an account on Quora to read this I believe.

https://www.quora.com/Why-do-deterministic-and-non-deterministic-Turing-machines-have-the-same-power

The basic idea is that a deterministic TM just executes ALL the possible paths that a nondeterministic TM might take. 

 

Edited April 8, 2019 by wtf

[Endy0816]

1 hour ago, wtf said:

Doesn't help. Same level of computability. Deterministic TMs do exactly the same things as nondeterministic ones and vice versa.

Nondeterministic TMs may be better in complexity but not in computability.

https://en.wikipedia.org/wiki/Non-deterministic_Turing_machine#Computational_equivalence_with_DTMs

Here's a nice Quora thread explaining why this is true. You need an account on Quora to read this I believe.

https://www.quora.com/Why-do-deterministic-and-non-deterministic-Turing-machines-have-the-same-power

The basic idea is that a deterministic TM just executes ALL the possible paths that a nondeterministic TM might take. 

 

 

I'm imagining having a simple separate program on a timer that could randomly alter the code, interpretation of that code and memory of the AI. Likewise with the right setup the AI could more deliberately alter parts of this itself to a degree. Trying to think if that would be enough or not though so it isn't a TM. Going to have to give this some thought.

[wtf]

1 hour ago, Endy0816 said:

 

I'm imagining having a simple separate program on a timer that could randomly alter the code, interpretation of that code and memory of the AI. Likewise with the right setup the AI could more deliberately alter parts of this itself to a degree. Trying to think if that would be enough or not though so it isn't a TM. Going to have to give this some thought.

Nobody has ever worked out a model of computation implementable by a human being that goes past the ability of a TM. The Church-Turing thesis expresses the fact that no such model exists. It's not a theorem, only a hypothesis. It's stood for 80 years without refutation.

https://en.wikipedia.org/wiki/Church–Turing_thesis

 

Edited April 8, 2019 by wtf

[internetcynic]

On 4/6/2019 at 11:43 PM, wtf said:

No that is false. Newtonian gravity, for example, is not computable. The jury is still out on whether quantum physics is computable. There are many easily human-conceivable problems that can never be computed by an algorithm. The Halting problem is one such, and there are many others. 

I  love statements that assert something is 'impossible'. Such statement are utter bollox. At best you coudl write "Within the extent and framework of current human knowledge, xxxxxx seems to be impossible".

Fact is I could post a long litany of things eminent scientists declared 'impossible', but happened anyway.

And this is because you CANNOT PREDICT WHAT NEW KNOWLEDGE, SCIENCE,  OR BRANCHES OF SCIENCE WE MAY UNCOVER IN THE FUTURE THAT WIlL MAKE THINGS WHICH ARE NOT POSSIBLE NOW, POSSIBLE AT SOME FUTURE DATE.

In  1890 you could not have predicted cars, airplanes, space rockets,  world war 1 and 2, electronics, sattelites,  mobile phones, television the internet , computers, heart surgery , genetic engineering, nanotechnology, lasers, missiles,  LED's, transistors, ......and on and on.

However, you CAN say from looking back at the past, that science fiction, which is usually written by blue sky thinkers,. has a tendancy to come true. Thats why all the Space tech used i n Star Trek in the 1960's is here now or undergoing development - phasers, cloaking devices, communicators, tricorders, impulse engines, warp drives, teleports.....

 

 

Edited April 9, 2019 by internetcynic

[Strange]

1 minute ago, internetcynic said:

Fact is I could post a long litany of things eminent scientists declared 'impossible', but happened anyway.

I bet you would find it a bit harder to post a list of things that mathematicians had proved impossible but were later found not to be. (I wouldn't be surprised if there were one or two, but I am not aware of any.)

[internetcynic]

5 minutes ago, Strange said:

I bet you would find it a bit harder to post a list of things that mathematicians had proved impossible but were later found not to be. (I wouldn't be surprised if there were one or two, but I am not aware of any.)

Thats because Maths isnt a science, its a religion. or an Art

[StringJunky]

2 minutes ago, internetcynic said:

Thats because Maths isnt a science, its a religion. or an Art

Art is maths with too many variables.

[internetcynic]

The wide eyed innocence of most peopel when it comes to AI is amazing.

I worked in IT all my life, i started as a Junior programmer in 1978 writing COBOL66 and FORTRAN77 before the Internet was devised, and worked up to 2007. I did a BSc in Computing Science specialisng in Expert Systems, or what is now called Machine learning

AI hasnt event got started. what you are seeing emerging is whats known as 'weak' Ai - systems that can only learn one task, like bricklaying or facial recognition, or spotting trends ins mountians of data . We have not yet got off the ground with 'strong' AI, that is, systems that learn and then apply that learning to an entirely new, unrelated problem, eg a self driving car that can then learn to fly an aircraft without new programming ,rules or hardware.

Now lets make an observation which, when you look closely is startlingly true: Science fiction has a habit of coming true. Look at, for example Star Trek, from the 1960's. Almost every bit of 'space' technology they had, is now in use, or under development, - communicators, tricorders, matter transportation, sub light drives, cloaking, even warp engines (the Alcubierre drive).

EVERY job that doesnt require human ingenuity CAN and WILL be replaced by a AI system or robot, and twenty years later, even that wont save your job. In the relentless pursuit of profit, capitalist companies have no compunction in discarding people, the ultimate despensible part of the machine.

This is going to present a massive problem for the planet, because it will eleiminate millions of jobs and create hundreds of millions of unemployed. So, society has limited options:

1. Create millions of NEW jobs so far no one has thought of, to redeploy the unemployed, and make jobs protected from AI or rbot takeover (never going to happen)

2. Accept that millions of people will never work again, and pay them a living wage for doing nothing, OR, accept that a 'full time job with a living wage' will only entail aboout 10 hours work a week, and all the remaining jobs have to be divided up the remainign jobs between the population .

3. Realise that having millions of people unemployed with no prospect of bettering themselves is a reciepe for civil unrest and its how warmongers like Hitler and Stalin rose to power

4. Cull the population down to fit the job market size. (and Id like to point out the UN Agenda30 plans to do exactly that - cull the human popualtion down to 1 billion people living in 50 megacities, and barred from owning houses, land, cars, and banned from the countryside )

Either of those options is going to be a tectonic change in the way humans live, and will rock this species far more than the invention of printing, the Industrial Revolution or the Information Age has done so far.

When you add to this the current progress of the competing plans such as Global Elite plan to create One World Government, the islamic plan to create the World Caliphate, the International Marxist plan to create a World Socialist Dystopia, then i believe that we are thus approaching Kardashevs Great Filter, at which we will fail.

Edited April 9, 2019 by internetcynic

[Prometheus]

11 minutes ago, internetcynic said:

However, you CAN say from looking back at the past, that science fiction, which is usually written by blue sky thinkers,. has a tendancy to come true. Thats why all the Space tech used i n Star Trek in the 1960's is here now or undergoing development - phasers, cloaking devices, communicators, tricorders, impulse engines, warp drives, teleports....

Appreciate the enthusiasm, but if Star Trek says we can travel faster than light and Einstein - backed up by theory and evidence - says you can't, who are you really going to take seriously?

[internetcynic]

16 minutes ago, Prometheus said:

Appreciate the enthusiasm, but if Star Trek says we can travel faster than light and Einstein - backed up by theory and evidence - says you can't, who are you really going to take seriously?

Ah well this is a perfect example.

First of all, Einstien didnt say yo ucant travel faster than light, all the equations do is go infinite at excatly c, each side of c the line gently slopes  off. So if yo uare created at above c, then the problem is SLOWING DOWN BELOW c.

The next problem is time. The arrow of tiem is determined by entropy.  entropy is governed by the Second Law of Thermodynamics, which itslef relies absolutely on two completely unproven, untested assumptions, that a) the universe is a closed system (Ther eis no proff it is) and b) the amoutn of mass-enerygy-information is fixed (of which ther is no proof).  The second Law may well be wrong for the Universe as a whole. If thats the case, Time may well be a local phenomena, creted as a consequence of spntaneous up collapses of wave functiosn , genertaing local time.

There is no such thing a Universal NOW. All NOWS''s are relative and may even not be dependant. Ther may not therfore even be a tie dimension, ther is NOTHING in any of the equations that demands a time dimension.

Which actually makes sense

To observe an nD object you have to observe it in nD+1 - to look round the back of a 3d box, you move thru D4 to get round the back.  Every dimension contians 1 axis - up/down. left/right back/forward, call them what you will A 3D object thus has 6( oir 3 if you call them axial) freedoms of movement.

 

So consider - we also have a freedom of movemnt in D4 - foprwards and back wards (in tie, or spatially, its the sae thing). The line in D4 represenst all possible NOWS in the 3D frame. We can, however only observe ONE planck frame inD4 at a tie. This flicking forwards thru the plank fraes is like flickign thru the fraes on a movie. You only see one at a time.  in D5 we have NO freedoms of movemnt, just one single point, that represents all possible NOWS in 4D.

 

Ill expand ore on this if you want, but now its time to cook a curry.......

Edited April 9, 2019 by internetcynic