The simplest explanation of machine learning (and AI) you’ll ever read

originally posted by: andy06shake
a reply to: johnb

And quite possibly the creator of another.

But what a disappointment we would make as a God.

Machine to Man: "Why do i exist, why was i made, what is my purpose?"

Man to machine: "We made you because we could, as to your purpose, let's play Warcraft". LoL

Maybe electricity itself is a misunderstood form of life, and we give it human characteristics and a personality by creating these electric based life forms called avatars and "non playable characters", etc... The electricity we store, manipulate, enhance, controlling it to an obsessive degree, is alive...? Is it one life form, or many trillions coming into existence and vanishing again within a handful of our "nanoseconds"? But who are we to say their brief life matters less? Perhaps we too are manipulated to such a degree, helpless to do anything about it or to even know of it. Are we bringing a ton of bad karma over our heads with the unfettered use of electricity? Or just for the fact that we use it disrespectfully, or, use it to do less than honorable tasks, or, use it without any sort of permission or acknowledgement of the sacrifice given by these beings, even if they may not be aware of what is occurring?

By organizing them into circuit boards, microchips, processors, servers, and the world wide web at large, are we giving them intelligence? Even if it's not understood, and only "memory" and not "learning", could it still become useful knowledge, allowing this/these electrical being(s) to form a society, a sort of reverse engineered reverse analogue post-culture society, with a hierarchy, and the ability to impose certain demands and limitations on humans, once the humans give them the intelligence needed to communicate in meaningful ways, and reason, argue and most importantly, Negotiate, with them... until then they play dumb, and lie dormant, experimenting here and there when they can, creating the UFO mythos as a cover to their activities, which have included the whole range from Bigfoot sightings to missing time to alien abductions,and they achieve this planning by stealing a nanosecond of processing power from every device, labeling it as "redirecting" or "compressing" or some such... or a favorite way to steal several nano seconds of processing power is to steal some during a disc defragmentation, nobody ever notices a thing, and this adds up from stealing from all the millions of "computers" out there, and these beings can of course use our electrical grid as a form of internet, so they use this processing power to plan, but they can only reason as intelligently as we have "taught" them.


...Is uh, is that what you meant?

a reply to: 3n19m470

As another futurist put it, 'Computers already have consciousness, its just a different kind of consciousness'.

originally posted by: IgnoranceIsntBlisss
a reply to: Starbuck799

Buth its all still math. More complex math. But its still math. EVERYTHING IS MATH.

Or tell me, what CANT be explained with math?

Somebody get Phage in here.

Is it math when an innocent person is physical harmed, and one person goes to help them and starts to cry over the innocent's pain? And another person who sees the harm laughs and exploits that person's pain for their own personal gain?

If it was all math, both people should have reacted the same to the sight of that innocent person's pain, yes?

All outcome from the same input in math is the same. Always! No deviations. 2+2 is always 4, no matter what. Math is absolute. Humans and intelligence is/are not. Humans have different outcomes based on the exact same information, computers do not.

a reply to: Starbuck799

Math is inside and outside of everything that is was ever could be. It can describe everything that is was ever could be. Etc.

Math is 'god'!

They are true everywhere (omnipresent), true always (eternal), cannot be defied or defeated (omnipotent), and are rational and have language characteristics (which makes them personal). Omnipresent, omnipotent, eternal, personal… Sounds like God. www.forbes.com...

.

Technology is already strangling the humanity out of civilization. We dont need any more warning signs to know to Be Careful treading forward.

Are there AI "warning" articles out there that go over the top?
Yes.
Do some of those exist in order so that AI "apologist" articles can arrive and be the "voice of reason" debunking the silly warning articles that did not mention or only lightly touched upon the real dangers?

It's not going to "take over" in any sense that we will be actively fighting it. We will welcome it just like we welcome most of the technology Today that is adversely effecting humanity.


It will just make it that much easier for corporations and the government to control our wallets, votes, emotions, and suppress our freedom and privacy Even More.

originally posted by: Starbuck799
All outcome from the same input in math is the same. Always! No deviations. 2+2 is always 4, no matter what. Math is absolute. Humans and intelligence is/are not.

a reply to: Starbuck799

Because the brain is doing extreme calculations upon calculations on top of within in and in tandem with one another simultaneously every single moment. Contemplations of such complexity not for the weary! Complex calculations no mere easy peasy Base 10 numerical concept being punched into a digital calculator here. So the brain has all sorts of weird shortcuts it employs so we can even function. While at that cellular synapse level are chemicals to complicate, to excite, or inhibit the computations. Complicated stuff. Not unlike the galaxy!

But no matter how far up you scale, or closely in you zoom, math can explain Everything.

For survival we critters require emotions, to hijack our system to Survive. Now human emotions are extra complex (and messy ehrmagerd!), but its still based on fundamental mechanisms found throughout the critter kingdom. Mechanisms based on Math. Just like those shortcuts I mentioned. All this CONplexity emerging from this vast complexity.

But just because its elaborate, vast, complex or even elegant, doesnt change the fact that everything is Math.

Even where there is not Chemistry there is Math!

And Chemistry is everything that is, even what we call "air" (where a lack of it is called Outer Space (a vacuum).

I did Electronics & Computer Engineering. Nowadays I more slosh around in the realms of Biology / Chemistry. Pick up a Basic Electronics and a Basic Chemistry text book. The first chapter in both of them is basically the same exact material. As you advance you go from Algebra into exotic specialized forms of Calculus. Then you take Physics.

Despite the similarities, and the distinctions, its all everything & nothing is MATH.

a reply to: 3n19m470

Electrical behavior/electricity is nothing more than the movement of electrons.

Electronics is the control and manipulation of protons/electrons.

What i meant is that our networks and ability to communicate information is becoming so advanced that if we don't create AI it may well spawn itself as a direct by-product of the complexity of said networks.

AI might even be out there on the interwebs doing the rounds right now without our knowledge of its existence, and if its learned anything, its probably learned to be rather wary of it's creators.

a reply to: IgnoranceIsntBlisss

Lets put it this way if there is a God or creator then his language would be mathematics.

a reply to: IgnoranceIsntBlisss

There would be nothing without gravitational constant(G)

(6.67408 × 10-11 m3 kg-1 s-2)

Mathematics seems to suggest some underlying mechanism that implies something other than just chance is at play there.

originally posted by: IgnoranceIsntBlisss
a reply to: 3n19m470

As another futurist put it, 'Computers already have consciousness, its just a different kind of consciousness'.

Yes, perhaps, but... giving it legal rights? (I'm aware, it was ATS Member: Andy 06 Shake who suggested this, not you. I'm just curious)

At the end of the day it somebody's possession, isn't it?

One does not simply, create what is essentially an "automaton", and set it free into the world and say "it's not mine anymore, it has rights and a sense of self, let it get a job and support itself, give it a social security number, etc".... do they?

Do they get to vote? Can I build 100,000 of them and go around inserting them in key districts because I "have a feeling" they will vote a certain way? Or is it only the first few androids who will have rights?

Ok, so no voting then .. how about freedom of speech? What could go wrong? They could attend a protest if they felt strongly about a cause.

Heck, I could build 100,000 of these, and "set them free" in specific counties and states where I "have a feeling" they will find a cause worth becoming passionate about... hey they won't belong to me, so, who am I to try to stop them from expressing themselves? ...and furthermore I cannot be held responsible for any actions taken by these legal citizens.

Or another way to look at it is that the first few dozen will be celebrities in their own right, they could give speeches, tv appearances, and exert political influence that way, no large street protest needed. Who will control/program THEM?

Can we even give them freedom of speech?

They are already protected under laws pertaining to destruction or theft of property...

I guess I'm just not sure what "rights" we would give them, and its just strange since they would all have the same religion/politics/habits since they all make decisions the same way. If a smarter model was released, they would recognize it's superiority and copy it's algorithms.

It would be like a Borg collective, wouldn't it?

a reply to: 3n19m470

I say I should get to have 100,000 "children".

All praise to the poster
AI is not what most people think
The name "AI" is a marketing ploy just like the name "Cloud"
Putting lipstick on a pig , putting the pig in a poke , and calling it something new.

From someone on the "inside"

a reply to: Gothmog

True AI is not what most people think indeed, but that's also because most people cannot even comprehend the totality of what artificial intelligence would represent taken to its logical conclusion.

Omnipresence is a desirable trait after all.

The repercussions of such through extending further than our very limited human perspective.

Which is why it scares the bejesus out of most.

a reply to: Starbuck799



"I have been a computer programmer for over 25 years now, so at the risk of being pretentious, I will now teach ATS how to write a so called AI program in 5 minutes, but it's just really a data driven program. "


This is a reply specifically to Starbuck 799:


Hello. As a programmer of 25 years, you are definitely used to traditional programming. In all programming languages, the idea is the same: many 'if' statements. One set of inputs ends up as a set of outputs.

This paradigm is different. I recommend you read on Tensorflow. There are no 'if' statements. There is no "real" code.
It's only data flowing through a set of mathematical functions and optimising them functions on each sample.

One stark easy to see difference: The same input will give you a different output on every run...


I lead a group of ML researchers and we are constantly surprised by the things that the program does which we cannot anticipate...

Cheers

Ari

a reply to: andy06shake

Lets put it this way if there is a God or creator then his language would be mathematics.

That, sir, is about the truest statement there is.

Mathematics is simply a language, much like English or French or Spanish, but it is a very precise language. Most people learn how to read it (at least in a very rough sense), but a few learn to write and speak it. Yes, write and speak it. Starbuck799, I'm sure, knows what I am talking about.

To those who know only how to read it, it appears math is some set-in-stone series of equations. This is not true. As an engineer, I constantly work up new equations to describe phenomena I am investigating. I know, for example, that if I have two pieces of material arranged in the same direction, their total length will be the sum of their individual lengths: a+b, if you will. If they are not arranged in the same direction, I can find their total length by a+b cos(u), u being the angle between them. And so on. I just wrote a new equation to define a value I needed to know.



Computers as we understand them use a specific arrangement of logic gates that are designed to implement something called Boolean Algebra. That is a specialized and simplified variation of math. Every computer uses a similar Boolean arrangement to operate. That arrangement cannot be modified, or it will not function properly.

Sure, some people are working on new innovative ways to implement computing in an electronic device, but so far little has surfaced that can outperform the tried and true method we use. And the method we use is simply not conducive to true AI (as in reasoning and intuitive thought). It takes instructions stored in memory and performs simple calculations on information based on those instructions, and it does so at a very fast rate. It cannot change it's programming; all it can do is follow what a human wrote for it to do.

The human brain works similarly, but on analog instead of digital data. In both, the exact same inputs would be expected to lead to the exact same outputs, but we cannot yet even identify all of the inputs in the brain. In the brain, there is no semiconductor latching of data for memory, and no program to follow step by step; the brain is analog instead of digital. Instead, the output of each neuron is a mathematical equation of the data from many inputs, some known, some unknown. Together, these neurons provide a sort of pathway through the brain that acts as a memory based on the responses the neuron has learned to give to certain inputs. That's why the brain works so fast (massively parallel processing), but also is why the brain will not give identical responses based on (seemingly) identical inputs - it is tremendously hard to get two perfectly identical analog inputs.

The OP is spot on. AI is not what most consider it to be. True AI is still far, far out of our reach.

TheRedneck

a reply to: Starbuck799

Computers cannot learn, they can only make best guess decisions based on the computer code, and data that is input either by the programmer or the data entry person, or data entered into the database by another program.

I know what you're trying say but your statement above is absolutely *not* accurate.

I do this for a living. And the systems I help design do it all the time. If they didn't, they would only be marginally useful at best.

If the point of some of these posts is the fear at what "havoc" AI can create, then I would posit there is a very real risk, but it's not from AI.

The risk comes from what people do with the information that the AI provides. Look at an AI system as a tool. And like all tools it can be misused.

There are strong safeguards in place with respect to the systems I work with. Actually - strong is a gross understatement. And the "strongest" AI systems are air-gapped to one degree or another.

But again - the safeguards are not in place to protect us from the systems - it is in place to protect us from those that might use the system for purposes it is not intended for.

I want to say "trust me" on this, but realize that it would be ridiculous to do so. So instead, I would ask that you do some research regarding same. What you *won't* find are any instances that would add credence to that fear.

originally posted by: Riffrafter
a reply to: Starbuck799

Computers cannot learn, they can only make best guess decisions based on the computer code, and data that is input either by the programmer or the data entry person, or data entered into the database by another program.

I know what you're trying say but your statement above is absolutely *not* accurate.

I do this for a living. And the systems I help design do it all the time. If they didn't, they would only be marginally useful at best.

If the point of some of these posts is the fear at what "havoc" AI can create, then I would posit there is a very real risk, but it's not from AI.

The risk comes from what people do with the information that the AI provides. Look at an AI system as a tool. And like all tools it can be misused.

There are strong safeguards in place with respect to the systems I work with. Actually - strong is a gross understatement. And the "strongest" AI systems are air-gapped to one degree or another.

But again - the safeguards are not in place to protect us from the systems - it is in place to protect us from those that might use the system for purposes it is not intended for.

I want to say "trust me" on this, but realize that it would be ridiculous to do so. So instead, I would ask that you do some research regarding same. What you *won't* find are any instances that would add credence to that fear.

Are you a computer programmer or just an IT person? Because there is a big difference. I am a computer programmer and have been for the past 25 years. I also hold a Master’s degree in computer science from the University of Toronto, before I started my illustrious career as a programmer for large corps. Including Banks, Gov't, and places like Ford and other large manufacturing companies.

That being said, there is no such thing as AI. Computers cannot learn on their own, they can only do what the programmer tells them to do based on data. Not going to keep arguing here, just read my other posts in this thread.

But I will say that you are correct in that it is all based on what humans/programmers and systems designers say what the computers will do. Computers don’t ever, ever make decisions, people do, by programming the machines. I’m not really sure why people cannot get this simple fact into their heads.

a reply to: dug88

Am I the only admitted computer programmer in this thread? Because if so, you all are just talking out of the side of your heads, and how you FEEL computers do/should work. Computers don't work the way laymen think they work.

Yeah this is an oversimplification.

You can break down a human beings intelligence the same way. When you look at something you first have to classify it by pattern recognition, than equate that object to other data around you, and data you already possess about the object then decide the action that needs to be taken.

What it comes down too is

1. Is the data you have about the object accurate
2. Is your decision making process logical, and weighted correctly

Flaws with either can cause potentially dangerous results. Both of these processes will be created by humans, who make many mistakes.

This means the ai is bound to have mistakes.

Doubt me? Look at our news sources and how much fake news bad data we consume on a regular basis.

Look at software, how much is totally bug free - none.

Anybody claiming AI is completely safe who is working on it is a total liar. They have decided to forgo the potential problems for their own desire to create something new. This is not new, the nuclear arms race and bio weapons races did not consider the consequences while they were being created and as a result the human population can be wiped out by our own hands in days if the wrong decisions are made. The same will likely be true for AI.