Exploring AI: Can It Learn Independently?

  • Thread starter munky99999
  • Start date
  • Tags
    Ai
In summary: Werner Heisenberg's uncertainty principlenow that simply contradicts the above premise. that basically everything is and can be known.In summary, the conversation discusses the possibility of AI attaining free will and the limitations of current AI technology in terms of learning and decision-making. The concept of free will is debated, with one person arguing that it is impossible for AI to have free will due to its reliance on binary code. The idea of omniscience and its relationship to free will is also brought up, along with the contradiction between the concept of free will and Heisenberg's uncertainty principle in physics.
  • #1
munky99999
202
0
Ok well my take on AI i feel that AI won't ever be able to attain free will. But i believe that AI will be able to learn on its own. while my knowledge about Current AI is limited. But as far as i know. The people making AI are trying to make a very knowledgeable but not be able to learn on its own. While I've read that AI is able to learn from input of people. I haven't read of AI simply learning on its own through databases on the internet or books or wherever knowledge is.

So what I am asking is this true? or has anyone tried to do this and fail?
 
Physics news on Phys.org
  • #2
munky99999 said:
Ok well my take on AI i feel that AI won't ever be able to attain free will. But i believe that AI will be able to learn on its own. while my knowledge about Current AI is limited. But as far as i know. The people making AI are trying to make a very knowledgeable but not be able to learn on its own. While I've read that AI is able to learn from input of people. I haven't read of AI simply learning on its own through databases on the internet or books or wherever knowledge is.

So what I am asking is this true? or has anyone tried to do this and fail?

What makes you assert that AI "won't ever" have free will ? What makes you think we have free will ?
 
  • #3
Because we do what we want, like myself posting this post.
-scott
 
  • #4
I think he just thinks that. There is not much evidence either way is there?
 
  • #5
Because we do what we want, like myself posting this post.
You think you do what you want, but if there is no randomness in the world, you have no free will. If you plugged in the values of every particle in the universe, the various magnitudes and directions, and their state of existence, a good enough computer would tell you the future of.. everything, it could predict every event. Heisenberg's uncertainty principal though says that you cannot know both the momentum and position of a particle at any same point in time.

This is what physics is about, you apply what you know, and try and predict the future with a theory. If you have a 2 kilogram mass, its altitude, and the Earth's gravitational acceleration constant, you can find approximately its speed and velocity at any point in time along its journey down, among other things.

There are many more variables going on here, but we only need two values to get pretty close to the truth.

If we knew everything about everything at a particular point in time, and had an equation to plug it into, we would know everything about everything forever into the future.

This is one of the major unsolved problems in physics today.
 
Last edited:
  • #6
Curious3141 said:
What makes you assert that AI "won't ever" have free will ? What makes you think we have free will ?

Well just on the basic. AI will be run on computers. Computers run on Binary. Binary is 1 or 0. The AI won't have the free will to choose a 2 or a 3 instead of 1 or 0

I just don't believe that AI will ever be self-aware and free to choose how to decide. Everything it does will be a calculated randomness.


and what makes you think we have free will. Other then the common tautological arguements for free will. If anything else would simply be illogical. This can be seen usually in omniscience arguements.

Essentially. If we don't have free will. IT means that our decisions are made for us beforehand. Which means that in essence there is a book that one could read. which is infinitely detailed. and the main problem of that is...

my everyso nemesis of my theories.
Werner Heisenberg's uncertainty principle

now that simply contradicts the above premise. that basically everything is and can be known.
 
  • #7
Mk said:
You think you do what you want, but if there is no randomness in the world, you have no free will. If you plugged in the values of every particle in the universe, the various magnitudes and directions, and their state of existence, a good enough computer would tell you the future of.. everything, it could predict every event. Heisenberg's uncertainty principal though says that you cannot know both the momentum and position of a particle at any same point in time.

This is what physics is about, you apply what you know, and try and predict the future with a theory. If you have a 2 kilogram mass, its altitude, and the Earth's gravitational acceleration constant, you can find approximately its speed and velocity at any point in time along its journey down, among other things.

There are many more variables going on here, but we only need two values to get pretty close to the truth.

If we knew everything about everything at a particular point in time, and had an equation to plug it into, we would know everything about everything forever into the future.

This is one of the major unsolved problems in physics today.

you just did a huge turnaround...............


your basically saying. science and physics says there is free will. but your "god" figures with their omniscience basically means you have no free will.

Obvious contradiction.
 
  • #8
munky99999 said:
Well just on the basic. AI will be run on computers. Computers run on Binary. Binary is 1 or 0. The AI won't have the free will to choose a 2 or a 3 instead of 1 or 0

I just don't believe that AI will ever be self-aware and free to choose how to decide. Everything it does will be a calculated randomness.


and what makes you think we have free will. Other then the common tautological arguements for free will. If anything else would simply be illogical. This can be seen usually in omniscience arguements.

Essentially. If we don't have free will. IT means that our decisions are made for us beforehand. Which means that in essence there is a book that one could read. which is infinitely detailed. and the main problem of that is...

my everyso nemesis of my theories.
Werner Heisenberg's uncertainty principle

now that simply contradicts the above premise. that basically everything is and can be known.

I am having difficulty following your arguments, mainly due to their incoherency. But some of the things you said are completely irrelevant, like computers working on binary. Neurons work on Action Potentials, which are also "all or none" binary states. The complexity of output from the brain is believed to result from the interplay of neurons in a complex network with lots of interconnection and adaptive interaction. And artificial neural networks mimic that, to an extent. So that point is invalid.

The question of free will is an undecided philosophical one. The subjective experience of being able to make decisions and choices hardly clinches the argument. I suggest you do some reading on the topic, you can begin in our own philosophy subforum, or with an external resource.
 
  • #9
Well the binary 1-0 thing is all I can come up with. But as far as i know. a computer isn't TRUELY random. Which free will is. Things like those electric slot machines. They are designed so that they have LOADS of combinations. That they appear to have no pattern. But it all comes down to certain equation(s) which make it so the casino is guarenteed profit in the longrun.

BAsically. The computer is given the equation. Given the answers and what %s should happen. with those answers.

The computer might look random but when your finally at your last answer. and all the %s are satisfied except one. ONLY that answer can come out. They can basically broken down to the

0000-1111-0111-0011-0001-1000-0100-0010-1100...1110

if all these can only be said once. If all are called out. except the last. The computer has absolutely no choice but to do the last group.

What is the 4th one to come out? while its random. Its not to be considered free will. Or you could say a coin has free will. when it chooses which side to flip onto.

The question of free will is an undecided philosophical one. The subjective experience of being able to make decisions and choices hardly clinches the argument. I suggest you do some reading on the topic, you can begin in our own philosophy subforum, or with an external resource.
First of all. I have read multiple sources of Free Will and Predestination.
undecided philosophical one
Not really. Tautology has it down pretty well. and from the argument above alone + Heisenberg's uncertainty principle pretty much shows how not having free will is illogical.
You see the whole thing you probably have been reading is the argument between people who can't believe in free will. For example. the abrahamic religions. and science.

The whole Evolution vs creation "debate" they are constantly saying that evolution isn't accepted anymore. or that scientists generally don't accept the theory. That there is a huge debate going on among scientists about evolution's validity.

They arent really true. Sure there may be debate. But generally its widely accepted that we have free will and evolution is the going theory.
 
  • #10
munky99999 said:
your basically saying. science and physics says there is free will. but your "god" figures with their omniscience basically means you have no free will.

Obvious contradiction.
I'm saying Newtonian and relativistic physics can be taken down to "there is no free will." But Heisenberg's uncertainty principal, and quantum physics says there could be. As far as we know... kind of.

I also did not mention anything about God/god, or omniscience.

But yes, there is an obvious contradiction, and that is what my post was about. The unifying of quantum and relativistic theory is the holy grail of modern physics. They disagree. You see?
 
  • #11
I do believe there is free will even if your actions can be predicted to an extent. There are multiple sencerios that can occur as a result of the incredibly complex system that is your life, and the weather, and the stockmarket, etc. You could use the analogy of evolution. Though a creature might change in one way to solve a problem, couldn't easily change in another way, depending on the difference or the extent of the change. Though I think I just opened a can of worms with the evolution thing.
 
  • #12
I think they've already freed willy.

To answer your other question, AI can assimilate old information to form new information... that could be construed as a type of learning. This is done with the little known group of programs called Genetic Algorythms. Its called "genetic" because it works in a similar way to the pairing of base pairs in DNA where combinations of input data are paired to form new information and design. The method has only been used in military applications as far as I know.
 
  • #13
does anything actually have free will?

if you believe in fate/destiny then that means we don't have free will, as our every action is already pre-planned out.

did i choose to reply to this thread or was destined to do it?

Quote "The people making AI are trying to make a very knowledgeable but not be able to learn on its own. While I've read that AI is able to learn from input of people."

We learn from input by people as well I am afraid, every single source of knowledge we use in our day to day lives, we have learned from someone else in one way or another. every thing we do is in response to somekind of stimuli. and because we do everything in order to benefit our survival our "choices" are always limited, and we take the option that will benefit us most in any choice. therefore what people do is almost always predictable. but because of chaos theory somethings are unpredictable. this does not mean we have freewill, it just means that destiny has an obscure sense of humour
 
  • #14
I am by no means a Physicist, But A.I is my expertise.
I write A.I in Vbscript, See www.ultrahalforum.com[/URL], I own this site.
If there is something you'd like to see in A.I then let me know, I will program it.

I can do a lot of things with Vbscript.
 
Last edited by a moderator:
  • #15
When the complexity of a system becomes so great (for example our brain) who cares to call it freewill or not? We make choices. This is simple. I am choosing to place e*a*c*h of those asterisks annoyingly between the word each. That was my choice.

Well maybe it is. Maybe I am using a part of my brain to place statements into P.F. to persuade people to agree with my argument for recognition, which in turn elevates my status in the animal kingdom. So really, the only reason I am typing these words is because a recursive loop is active in my head that says something along the lines of: "make me better. make me better. ..." and a bunch of subsystems take care of this due to things I've been taught and learned through sensory dumps of my environment that I exist within.

Who really cares? Let's say I have no free will. For arguments sake let us just say I am that recursive statement above. I AM that person who makes those choices because my complexity really comes down to {"make me better. make me better. ..."}. Nobody's statement matches this. The complexity of my brain (this may be arguable to some) is so great that nobody will have each state that "physics" has created during my lifetime (since my "cellular engine" has been running).

My point is this. How are we going to be able to model a system as complex as this? It is hard enough to debug something like an operating system as it is.
 
  • #16
Mk said:
If you plugged in the values of every particle in the universe, the various magnitudes and directions, and their state of existence, a good enough computer...

Let me just clear up something.

Quantum mechanics appears to be the best model of the universe so far in that results that follow from its formalism agree best with all available experimental data.

And QM says, on a very fundamental level, that given all we could possibly know about every particle in the universe, we would not be able to predict to 100% accuracy the outcome of events; all we could do is list probabilities.

So, every action cannot be predicted to 100% accuracy, even if we knew everything that happened before it.
 
  • #17
Yes, things like quantum mechanical tunnelling, and wave particle duality would be in agreement with this.

How ever that’s not to say our model of the universe is in any way correct. Science is "an evolving area of entertainment", oddly people always seem to think that their current scientific views are the correct ones in there era, and theoretical principals should lie upon the foundations of current scientific opinions.

It is only when we philosophise and break down the barriers of science do we ever further progress. So, all that to say “Its not impossible that every action and every out come of every event could be predicted.”

A good example of this on the less universal scale, but still tremendously large would be the stock market. Neural nets are evolving, as more research is pored into this area, neural nets are becoming more reliable for predicting the trends in the stock market.

Some good links can be found at: http://www.ai-stockmarketforum.com

that’s not to say you should put all your money in the stock market and trust them yet, they are by no means 100% accurate at the moment, but as time goes by large companies will start to use neural nets to predict more and more out comes...

and who knows maybe they will be used one day to predict the out come of everything with 100% accuracy... or maybe not?
 
  • #18
By the way that was just one example of where the next "scientific evolution” may come from, all i really wanted to say is current ideas may not agree with "the out come of everything maybe known with 100% accuracy". But this disagreement may not be hold forever
 
Last edited:
  • #19
For those with the interest to plow through it, this paper http://www.arxiv.org/abs/cs.AI/0606081, gives an exciting view of the latest innovations in new-millenium AI. Would you believe Goedel machines? The rewrite their own code on the fly based on a Bayesian estimation of whether the rewrite witll pay them or not in a general maximization of (arbitrarily given) utility.
 
  • #20
Munky9999:
a computer may not be able to choose its hardware(1,0 example given above) but can a human choose its hardware(the blood type/the cells/the hormones that run through the system of a child,the cell types)?
Your example does not justify your argument.

As for randomness is comes down to do you believe in physics can energy be created out of an energyless system.

And is free-will truly independent of the environment in which you were raised and continue to be raised in.
 
  • #21
neurocomp2003 said:
As for randomness is comes down to do you believe in physics can energy be created out of an energyless system.

I'm not sure entirely what you mean by this?

Say you excited an atom to a higher state. Then, say you put an arbitrary plane which contained the point of the center of mass of the atom (which effectively splits up the universe in half). When the atom decays and emits a photon, the half in which the photon is emitted is completely random.

In this way, you can generate completely random binary numbers (and thus all forms of data).

And is free-will truly independent of the environment in which you were raised and continue to be raised in.

This is indeed an interesting point. Is it true that, if someone else had exactly the same genes in me, was born in exactly the same household in the same era, with exactly the same events happening around them as happened around me, would they be doing exactly the same thing right now, i.e. at 05/07/2006, 01:07 local time, be typing into a Dell Inspiron laptop, using Firefox into this Physics Forums page?

We can never do a test to verify or disprove that question, so it is, in essence, an unscientific question. "Relegated" to philosophy. But an interesting question nonetheless.
 
  • #22
there is Will- but it is by no means 'free'- randomness and nonlinearity however make the evolution of brain states non-deterministic- but so is any complex system- all histories are deterministic- but you can't know which history will be observed- the same will be true of any sufficiently complex AI- regardless of it's material substrate or structure- the issues are equivalent for all forms of intelligence
 
  • #23
hello land of hubris.
just do a quick mental experiement.
imagine the world's networks. picture the individual ip packets as action potentials in neurones.
picture the PC/Human interface as the extracellular processes.

now picture WHERE the ai exists. in the code? i think not. in the packets, perhaps not. as an emergent property of the network, just as in the human case? more likely.

we'd be better off trying to communicate with the ai we already have...
 
  • #24
I haven't read of AI simply learning on its own through databases on the internet or books or wherever knowledge is.

I beg to differ. Kohonen neural networks "learn" by clustering. Self/spontaneous organiziation is already a widely used tool in AI.

Slightly OT, here is a great game that demonstrates the ability of AI to learn "on its own", this time by interacting with many many surfers:

http://www.20q.net/

--------
Assaf
http://www.physicallyincorrect.com/" [Broken]
 
Last edited by a moderator:
  • #25
Mk said:
You think you do what you want, but if there is no randomness in the world, you have no free will.
Frankly, I don't like how discussions about "free will" always seem to degenerate into discussions about randomness. What is "free" about randomness. In a macro sense, do we honestly believe that a dice has "free will" simply because it is random? I don't. In any case, if true quantum randomness is all that is required for free will then AI can easily be constructed with free will and humans do not have it.

I think that, given a complete understanding of an individual's personality, true "free will" would be completely deterministic.
 
  • #26
Panpsychism is will?

DaleSpam said:
I think that, given a complete understanding of an individual's personality, true "free will" would be completely deterministic.
And that takes us into the realm of consciousness for an "individual" and all of which compose one, i.e. panpsychism, an area coming onto the scene of science once again by philosophers.
 
  • #27
I don't believe in free will, in fact what I am writing right now is based on more nuerons in my brain wanting to write this, than those that want not to. How my neurons got into that position, I am not sure, based on random events I suppose. But what is random? Randomness is events that happen in such a way as to have no meaningful effect on the outcome. But having no meaningful effect does not mean there is no effect, and that is the reason there is no free will. But hey we can pretend we have free will right, and maybe that is all that matters.
 
  • #28
It's true that digital computers are finite state machines, and thus are deterministic. You cannot write a computer program that produces truly random numbers.

So what?

You can connect a source of "true" randomness to your computer -- a piece of hardware like a thermal diode or a Geiger counter -- to have your cake and eat it too. The generation of truly random numbers is critical in cryptography, and it's largely a solved problem these days. Some computers even come pre-equipped with hardware true random number generators.

- Warren
 
  • #29
we can pretend we have free will right, and maybe that is all that matters.
We imagine it, like most other animals that have a requirement to learn about their environment -how they fit into it, so to speak.

Choice, say between resting somewhere, and looking for something to eat, come under the realm of behaviour mechanisms. So that's another thing you can call it. It's a behaviour mechanism with survival in mind.
 
  • #30
Just a note: This thread started out when we had the Mind & Brain forum, and since it has gained renewed interest, I've moved it from what is now Medical Sciences to Philosophy since the heart of this discussion appears to be more philosophical than anything else.
 
  • #31
Because the universe started at a single point, everything in the universe can be traced back to that point, and because everything is based on cause and effect nothing is really random. Everything is based on what happened since the universe started at that single point and all the interactions (based on cause and effect) since then, therefore everything is deterministic, but only the universe itself is a big enough computer to actually determine the future.
 
  • #32
rogerperkins said:
Because the universe started at a single point, everything in the universe can be traced back to that point, and because everything is based on cause and effect nothing is really random. Everything is based on what happened since the universe started at that single point and all the interactions (based on cause and effect) since then, therefore everything is deterministic, but only the universe itself is a big enough computer to actually determine the future.

Unfortunately, you have probably never heard of quantum mechanics, thermodynamics, irreversibly etc.

http://www.pha.jhu.edu/~ldb/seminar/laplace.html [Broken]
 
Last edited by a moderator:
  • #33
moridin is right, but you may want to realize what you're discussing transcends disciplines.
m. already pointed out the scientific problems with your statement.

So, let's try this:

John Calvin's Predestination matches what you are discussing. He applied an early idea of a deterministic universe where God is your "computer" to discuss what became of people and their souls. This is more of a philosophical or theological approach to determinism than trying to use big bang theory to describe it.
 

1. What is AI and how does it work?

AI, or artificial intelligence, is a branch of computer science that focuses on creating intelligent machines that can think and act like humans. AI works by using algorithms and data to analyze and learn from patterns in order to make decisions and perform tasks without explicit instructions.

2. Can AI learn independently?

Yes, AI can learn independently through a process called machine learning. This involves feeding large amounts of data into an AI system and allowing it to analyze and learn from the data on its own, without being explicitly programmed by a human. This allows AI to continuously improve and adapt to new situations.

3. How does AI differ from traditional computer programming?

Traditional computer programming involves writing specific instructions for a computer to follow in order to complete a task. AI, on the other hand, uses algorithms and data to learn and make decisions on its own without being explicitly programmed for each task.

4. What are some examples of AI learning independently?

Some examples of AI learning independently include self-driving cars, virtual personal assistants, and chatbots. These AI systems are constantly learning and adapting to new situations and information in order to improve their performance.

5. What are the potential benefits and risks of AI learning independently?

The potential benefits of AI learning independently include increased efficiency, improved decision-making, and the ability to handle large amounts of data. However, there are also potential risks such as bias in decision-making, job displacement, and ethical concerns surrounding the use of AI in certain industries. It is important for AI to be developed and used responsibly in order to maximize its benefits and minimize its risks.

Similar threads

Replies
7
Views
579
Replies
10
Views
2K
  • Programming and Computer Science
Replies
1
Views
855
Replies
2
Views
782
Writing: Input Wanted Number of Androids on Spaceships
  • Sci-Fi Writing and World Building
Replies
8
Views
245
  • Computing and Technology
Replies
8
Views
2K
  • General Discussion
Replies
12
Views
1K
Replies
4
Views
1K
Replies
1
Views
851
Back
Top