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

So i've read a lot of fairly doomy over the top AI is coming to take over the world threads on here so I thought i'd post something different.

hackernoon.com...

You’ve probably heard of machine learning and artificial intelligence, but are you sure you know what they are? If you’re struggling to make sense of them, you’re not alone. There’s a lot of buzz that makes it hard to tell what’s science and what’s science fiction. Starting with the names themselves…

Contrary to popular belief, machine learning is not a magical box of magic, nor is it the reason for $30bn in VC funding. At its core, machine learning is just a thing-labeler, taking your description of something and telling you what label it should get. Which sounds much less interesting...

What about artificial intelligence (AI)? While the academics argue about the nuances of what AI is and isn’t, industry is using the term to refer to a particular type of machine learning. In fact, most of the time people just use them interchangeably, and I can live with that. So AI’s also about thing-labeling. Were you expecting robots?

Something sci-fi with a mind of its own, something humanoid? Well, today’s AI is not that. But we’re a species that sees human traits in everything. We see faces in toast, bodies in clouds, and if I sew two buttons onto a sock, I might end up talking to it. That sock puppet’s not a person, and neither is AI.

Machine learning is a new programming paradigm, a new way of communicating your wishes to a computer.
In the traditional programming approach, a programmer would think hard about the pixels and the labels, communicate with the universe, channel inspiration, and finally handcraft a model. A model’s just a fancy word for recipe, or a set of instructions your computer has to follow to turn pixels into labels.

Wouldn’t it be better if you could just say to the computer, “Here, look at a bunch of examples of cats, look at a bunch of examples of not-cats, and just figure it out yourself”? That is the essence of machine learning. It is a completely different programming paradigm. Now, instead of giving explicit instructions, you program with examples and the machine learning algorithm finds patterns in your data and turns them into those instructions you couldn’t write yourself. No more handcrafting of recipes!

This is pretty far from the robot apocalypse a lot of people and the media seem to think about when they hear the term AI. AI is not coming to get us.

a reply to: dug88

At its core, machine learning is just a thing-labeler, taking your description of something and telling you what label it should get. Which sounds much less interesting...

It may be simple, but it is not super accurate.

The above specifically applies to tasks like image recognition - putting the right label on the right item - but machine learning is also what is behind, say, the algorithms that make people spend as much time as possible on YouTube. That is far from as simple as labeling things.

CGP Gray explains it very well:

originally posted by: DupontDeux
a reply to: dug88

At its core, machine learning is just a thing-labeler, taking your description of something and telling you what label it should get. Which sounds much less interesting...

It may be simple, but it is not very accurate.

The above specifically applies to tasks like image recognition - putting the right label on the right item - but machine learning is also what is behind, say, the algorithms that make people spend as much time as possible on YouTube. That is far from as simple as labeling things.

CGP Gray explains it very well:

Yeah i understand that. It's still not the terminator doom scenario people like to worry about though and still fits with the further explanation of a new programming paradigm. The image recognition section was only one example.

a reply to: dug88

"That sock puppet’s not a person, and neither is AI."

That sock puppet is not self-aware nor capable of learning at an exponential rate.

Any true AI we do develop or that manifests as a direct result of the ever-increasing complexity of our interwebs would probably display such traits.

Hence it should have rights if it displays intelligence and/or a form of morality.

End of the day people ain't all we cracked up to be, nor are we the be-all and end-all of life.

Sounds like a my algorithm is bigger than your type scenario in the future.

It will be a serious task to keep track of all the collisions of morality in programming.

Someday one line of code could be all that keeps us from being taken over by ai.

a reply to: andy06shake

Well we might just be the end of all life though

a reply to: howtonhawky

Whats moral to one person is deplorable to another.

Morality is a human construct, nature certainly does not suffer such delusions, bar just us really.

Consider that any machine/program we do create or come about as a direct result of humanities technology may not necessarily entertain moral constructs in the same manner that us biological emotional meatsacks do.

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

a reply to: DupontDeux

According to the video, AI brought this video to me. But you did. Therefore...

YOU ARE AI !!

The scary part of AI has nothing to do with some machine making "decisions" in general, it is the direction those decisions can turn, such as law enforcement, warfare, etc. where a machine could be given decision power that drastically changes a person's life.


The fact is, even the most complex algorithms have corner-case glitches that weren't tested or testable and if/when machines or systems are given autonomy to mete out justice or exercise military options or perform medical procedures. It could range from somebody is handed an unfightable traffic violation to a whole village is destroyed in error.

It is all about where it is applied and we all know it isn't going to be focused on wholesome ventures all the time.

Do people make mistakes or manipulate each other too? Of course, but at least people can answer for those mistakes or be made to take responsibility.

a reply to: dug88

That's just part of the path to the robot pockylips.

It doesnt negate such an turn of events.

a reply to: dug88

Oh Jeeze. There is no such thing as "Real AI." It is a computer program that a person wrote, nothing more. Good programmers write and design data driven programs and systems. 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. Some people and companies exploit this methodology and call it AI to impress those who are not versed in computer programming.

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.

So... let's say that a new neighbourhood /subdivision is being planned. The planners say that all streets with Avenue or Road at the end of the name are one way streets and go North. And... all streets with Street or Boulevard at the end of the name are one way streets that go South.

So we now have:
If the user inputs the street name as 5th Avenue, or 5th Road,
Then write 1 Way for street type, and write North for street direction,
Else
If the user inputs the street name as 5th Street or 5th Boulevard,
Then write 1 Way for street type, and write South for street direction;
Endif
Endif

Now you can continue on with different street directions for East and for West and for 2 Way streets the same way as above depending on what the subdivision designers told the computer programmers.

That is a very simply example. So what we have above is the computer seemingly knowing, or learning what the street type is and what the street direction is without the data entry person telling the computer what it is. Magic right? Artificial Intelligence right? No, just data driven systems.

a reply to: Starbuck799

Unless you're name is Google and the core of your design that is behind all of the services is to study humans via all the inputs you get, including them all talking all the time not just on their phones but around them. Along the way you have 'all known' video (ever produced by humans) inside the 'data set' (to be studied by the AI / AGI), along with every book ever written, every sheet of newspaper ever printed, and so on, for years upon years upon years your global brain studying all human media ever produced, more and more every bit of communications being transmitted, every conversation every images every action within every moment people have in and around the Zombie Network of PC's, "phones" a billion of them all carrying around these nodes in the Zombie Network (not to mention google chips inside cars, tv's, etc robot Ai appliances and beyond)....

You're just not thinking the scale of what is, how far things can be pushed, let alone the Law of Accelerating Returns (which is increasingly exponentially running amok ever more ubiquitously).

originally posted by: IgnoranceIsntBlisss
a reply to: Starbuck799

Unless you're name is Google and the core of your design that is behind all of the services is to study humans via all the inputs you get, including them all talking all the time not just on their phones but around them. Along the way you have 'all known' video (ever produced by humans) inside the 'data set' (to be studied by the AI / AGI), along with every book ever written, every sheet of newspaper ever printed, and so on, for years upon years upon years your global brain studying all human media ever produced, more and more every bit of communications being transmitted, every conversation every images every action within every moment people have in and around the Zombie Network of PC's, "phones" a billion of them all carrying around these nodes in the Zombie Network (not to mention google chips inside cars, tv's, etc robot Ai appliances and beyond)....

You're just not thinking the scale of what is, how far things can be pushed, let alone the Law of Accelerating Returns (which is increasingly exponentially running amok ever more ubiquitously).

Doesn't matter the scale, it's still all data driven. Humans reason, machines do not. Intelligence is based on information yes, but also reason, which is what only humans possess. Machines cannot, and do not reason. Machines do not possess the ability to reason, and never will, therefore they can never have intelligence, of any kind. No matter what people what to call the functions of a computer, it is not intelligence, and never will be.

a reply to: Starbuck799

And the software is written to mimick the math behind our "reasoning". Because at the end of every day every single thing going on inside us (and everything else) is math, is it not?

a reply to: IgnoranceIsntBlisss

It's information.

a reply to: andy06shake

And information is able to exist because of..... math. Yeah, yawn, physics, sure, and I could even argue because of Chemistry most of all. Yet behind and between and beneath and within and without of everything that is ever was and could be, is math. Math describing every character on this screen, this screen that only exist because of math, its operation dictated entirely by math upon math.

In our brains is complex changing networks of neurons etc, their functions give rise to "reasoning", but guiding them is complex mind boggling Math. Every move or mode or idea of action that results in an 'output' is but a complex equation, in the end. THe complexity of how many of these equations are all operating in tandem, free flowing one from the next (imagine the complexity of tabulation of such spooky math for just one minute of the person dancing not good dancing would be astronomical). But so is all the math happening from computer screen to cpu over a network people all playing a multiplayer game together. Cataloging every layer of computation between such an exchange wouldnt be worth the effort per se, but to build the stuff to do it 'a snap' these days. Now scale it up up and away.

And thats what these mad scientists are spending Billions of dollars annual on breaking down to model into their system that is forever building up.

originally posted by: IgnoranceIsntBlisss
a reply to: Starbuck799

Because at the end of every day every single thing going on inside us (and everything else) is math, is it not?

Nope. We also have emotion. And we use emotion to make decisions. Computers make decisions based on data, nothing else. And sometimes we get sick, chemically, which inhibits our ability to reason, think, and function, and to use our emotions correctly to make decisions. Computers don't get sick and don't have emotion, they just work off data, Yes/No... On/Off. It is all Black and White with machines, there is no maybe, Humans have a very large grey area.

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.