is flipping a coin AI?
 

What about magic eight ball? Is that machine learning?

Yes and no

It depends on the actual usage.

It's AI if :

* You and your friend flip a coin to decide who is smartest.

* You got a good price on the magic eight ball because you used wikibuy price compare.

:1orglaugh

Definitely sure! :stoned

no, AI cant predict randomness. i thought about it too and read big article about AI predicting lotto. but no random is really random nobody can predict it/

edit: you may get correct answer sometimes, but it just pure luck, because coin has only two sides, but it gets difficult if there are more choices/options to choose from.

if you feed it data then it can predict stock market., i guess, some people did some research on it too.

Oxymoron. If random isn't really random then it should be able to be predicted.

With a coin toss there's nothing random about it, It's only random to us because we don't have the mental capacity or senses to do the calculations required in the amount of time given.

Actually it can and it does it much better than any overage human.

I do remember a program. Hmm... a program of about 80 bytes length (yes, just 80 or so bytes of code) made for a pocket calculator like this: Bz-34.

...and it was predicting the human seance of random numbers with up to about 90%.

looks like i was reading wrong articles.

The most interesting thing is that 80-byte program (its code was about two times shorter than a twitter message) was self educating. It was giving better and better predictions after each run. The simplest way to trick it was... guess which one? To use a real coin. Just stop generating the "random" numbers in your head and trust to real coin - the program will be defeated.

By the way, do you know what a nerdy Soviet kid like me has wanted to have when he was 12-13 year old? No, not a real semi-automatic gun, because I had it already. And not a woman, because I was unable to afford one (but I had a very good imagination ;)).

I wanted a programming calculator - a very expensive piece of Soviet shit which was able to run a 105-byte program and has a memory on 15 numbers.... including (wow!) the real ones with (wow again!) a floating point!

The first program I wrote was a 20-25 byte ballistic emulator of a tank fight. The player (a cannon operator) had a few attempts to hit a moving tank by changing the barrel angle till it (the enemy tank) enters its fire range (the range of its own cannon) and destroy you. The second game was a Moon landing simulator about 40-45 bytes length.

The 3rd program was a serial maniac emulator (LOL!) The program knew your exact location in X-Y coordinates, the location of a maniac and the location of a safe place where you had to arrive. Since it was a night you didn't know shit. The program just told you the distance (not coordinates) of the safe place and the maniac. So you were need to do a blind escape. Just choose the angle (from 0 to 359 degrees) and step in that direction. The maniac was moving about two times slower than you, but... he knew your actual location (a night vision device?) and he has never lost the track. Sounds like a very stupid game scenario based on Pythagorean theorem, but it gave me some feeling of a horror movie during cold winter nights in 80s ;)

And you can believe me or not, but all those programs were using real physical formulas to calculate things like a ballistic trajectory (well, excepting a windage). The Moon landing game was absolutely realistic, because it used the exactly same formulas that were used for the World's first soft Moon landing which was done by the Soviet spaceship "Luna-9" in 1966.

i dont understand most of the things you said.

but looks like your childhood was interesting.

i have played moon landing game in my childhood.

but never programmed on calculator or stuff like that, my first computer was Intel P2 equivalent i guess. i remmber this becoz everyone in my neighborhood was asking is it P2 ? is it P2 ?

and i said its not P2 , but its equivalent. or some lower model ., i guess.

It was a Soviet childhood so to be honest it sucked monkey balls. The electronics was a rare and very expensive thing. When I was a 12 year old child and passed the assessment exam at biology with "5" (an equivalent of "A" in the US school system), my mom bought me a super duper calculator for almost her monthly salary and I was making the "computer games on it". The force of gravity on the Moon is 1.62. The exhaust flow rate and a specific impulse for a rocket? No problem, Sir! I had to put the impulse parameters like the amount of the fuel to be burned out at the specific time and after a minute of counting (yes, it was about minute to execute a simple code), the calculator told me the current height, the speed, the amount of fuel and oxygen to breathe. There was no graphics and no animation. Only these numbers I saw on the screen. Yes, a very interesting childhood for a 12 year old kid :) This is the real toys the Soviet kid could afford... But hold on, I had a .22LR semi-automatic pistol - a real nice toy for a child.

This is my first computer, which I used to learn coding and to spend hours in space traveling, strategic war games, landing aircrafts, control the bathyscaphes and airships:

https://upload.wikimedia.org/wikiped...nika_MK-61.jpg

I even played various adventure games like "The Never Ending Story" on a piece of shit with a 105 byte memory. Bytes, not megabytes and not kilobytes. It was only able to remember 105 simple commands, like "plus", "minus", "sinus", "if.. then", "go to", "square" etc ("1" or "2" were TWO commands).

NO !

For any computed device to be considered Artificially Intelligent, It must be able to recognize a mistake and reprogram itself (change it's own code) for correction.

Anything less is not AI.
There have been plenty of programs that give that appearance, but do not actually contain that. Just branches of code to a possible result of input.

OP, I know you're just 'fucking around' but to humor your curiosity, MACHINE LEARNING is all about picking up on certain DATA SET PATTERNS (eg., what links do visitors to your sites click?) and making certain predictions.

If these predictions turn out to be right and fit your users' behavior, the rest of your site is optimized to convert as much of that traffic into COLD HARD CASH

This applies both to ON SITE and OFFSITE behavior

Flipping a coin without any predictive and behavior optimizing elements is NOT ML

ML and AI are not the same thing.
A machine can be programed to learn without reprogramming itself.
Facial recognition captures would be a fair example.
You can catalog these mathematical snapshots and search for a match automatically but it is not AI but ML.

ML-Learning is the capturing of input if you make decision on it or not but even if you do make decision, your branches of output are finite to human programming.
AI could self generate new branches of output on its own and not programed branch/choice. It may learn later it did not make the correct choice and change itself so it does not do the same again. Without human intervention.
Even Watson is not there yet. But gives that appearance. It is just using fuzzy logic and presenting the best possible result, but has to be told it was wrong by human reprogramming.