What we now know about Magic... Sorta...

  • Astrodragon
  • Astrodragon's Avatar
  • Away
  • Author
  • Author
More
4 years 7 months ago #9767 by Astrodragon
Astrodragon replied the topic: What we now know about Magic... Sorta...
Remember, traditional magic is all done in the mind. The rituals and tools are just props to help you get it right.

I love watching their innocent little faces smiling happily as they trip gaily down the garden path, before finding the pit with the rusty spikes.
The following user(s) said Thank You: Malady

Please Log in to join the conversation.

More
4 years 7 months ago #9774 by Phoenix Spiritus
Phoenix Spiritus replied the topic: What we now know about Magic... Sorta...

Arcanist Lupus wrote:

Tom Lehrer wrote: But in the new [math] approach, as you know, the important thing is to understand what you're doing, rather than to get the right answer.


In the modern world this is actually correct.

Computers these days are much faster and more correct then any human can be at number crunching, the trick is to know what to ask the computer to do. Converting a complex problem into a mathematical problem a computer can solve is what needs to be taught, the drudge work of mindlessly calculating sums is pointless in the modern world of everyone having a super computer in their pocket.
The following user(s) said Thank You: Malady

Please Log in to join the conversation.

More
4 years 7 months ago #9784 by annachie
annachie replied the topic: What we now know about Magic... Sorta...

Please Log in to join the conversation.

More
4 years 7 months ago #9789 by Phoenix Spiritus
Phoenix Spiritus replied the topic: What we now know about Magic... Sorta...
What everyone seems to not understand is that current computers are not thinking machines, they are calculating machines. The thinking still has to be done by a programmer, and the computer can only do as much as its programmer imagined for it to do.

Please Log in to join the conversation.

More
4 years 7 months ago #9792 by Arcanist Lupus
Arcanist Lupus replied the topic: What we now know about Magic... Sorta...
But when does calculating become thinking? As of four days ago, Computers have triumphed over Go Grandmasters. Go can't be mastered by brute force calculation. Is the computer thinking? If not, what is it doing, and how is that different from what a grandmaster does?

"Shared pain is lessened; shared joy, increased — thus do we refute entropy." - Spider Robinson

Please Log in to join the conversation.

More
4 years 7 months ago #9824 by Kristin Darken
Kristin Darken replied the topic: What we now know about Magic... Sorta...
Go and chess have end game scenarios. When certain patterns of pieces exist, it is possible to predict a path to completion. From there, you simply have to create an understanding of how THOSE patterns can be created... ie what is the end game sequence if you consider the known pattern that leads to end game as an end game in itself.

While the idea that it cannot be possible to brute force an entire game of GO, it IS possible to brute force an endgame and it is possible to program smart mid-game moves that lead to the development of patterns which can become end game leading.

How is that different than a Go grandmaster? it isn't, really. But the Go grandmaster can likely abstract the play of a game of Go and apply that understanding of strategy in other aspects of life. A computer doesn't even know that there is strategy involved, and has no way of intepretting outside events a if those events were a game of Go.

Fate guard you and grant you a Light to brighten your Way.

Please Log in to join the conversation.

More
4 years 7 months ago #9828 by Valentine
Valentine replied the topic: What we now know about Magic... Sorta...

Arcanist Lupus wrote: But when does calculating become thinking? As of four days ago, Computers have triumphed over Go Grandmasters. Go can't be mastered by brute force calculation. Is the computer thinking? If not, what is it doing, and how is that different from what a grandmaster does?

AlphaGo works in two parts. When it is the computer’s turn, the program first suggests moves based on the sorts of general tactics that human players have used in the past—much as Deep Blue would. Then the second part of the system sifts those moves for those that look like they might lead to a win, again based on patterns it has picked up through memorising zillions of previous games.


Since I doubt the Grandmaster has memorized "zillions" of previous games, the computer is not doing what the Grandmaster does. It may not be brute forcing the entire game, but it is brute forcing the moves that it checks. Until that computer can take what it "learns" from playing Go and apply it to a more complex situation, then it is still just a computer program.

So if you took the coding for Deep Blue's chess and programmed in the rules of Go, then Deep Blue became a competitive and eventually great Go player, that would imply intelligence.

Don't Drick and Drive.

Please Log in to join the conversation.

More
4 years 7 months ago #9830 by Kristin Darken
Kristin Darken replied the topic: What we now know about Magic... Sorta...

Valentine wrote: Since I doubt the Grandmaster has memorized "zillions" of previous games, the computer is not doing what the Grandmaster does.

Actually, it's not a zillion games of memorization that matters. Its the recognition that every game of Go is made up of a number of smaller games of Go. And each of those smaller games is also made up of smaller games of Go. And within the context of every large game of Go there are a number of smaller pieces of that game that can reflect an 'end game' pattern at any moment. Each of these smaller games are interwoven in such a way that to place one in final winning state can trigger the correct or incorrect completion to adjacent smaller games.

So you don't need to memorize zillions of previous games... you only need to observe the completion of one smaller piece and integrate that completion into the completion pattern of its neighbor... and the one beyond that... and so on, until your combination of patterns has completed the entirety of your gameboard.

Fate guard you and grant you a Light to brighten your Way.
The following user(s) said Thank You: Dreamer

Please Log in to join the conversation.

More
4 years 7 months ago #9832 by Valentine
Valentine replied the topic: What we now know about Magic... Sorta...

Kristin Darken wrote:

Valentine wrote: Since I doubt the Grandmaster has memorized "zillions" of previous games, the computer is not doing what the Grandmaster does.

Actually, it's not a zillion games of memorization that matters. Its the recognition that every game of Go is made up of a number of smaller games of Go. And each of those smaller games is also made up of smaller games of Go. And within the context of every large game of Go there are a number of smaller pieces of that game that can reflect an 'end game' pattern at any moment. Each of these smaller games are interwoven in such a way that to place one in final winning state can trigger the correct or incorrect completion to adjacent smaller games.

So you don't need to memorize zillions of previous games... you only need to observe the completion of one smaller piece and integrate that completion into the completion pattern of its neighbor... and the one beyond that... and so on, until your combination of patterns has completed the entirety of your gameboard.


I agree, but that isn't what the computer is doing. According to the article the computer is comparing "zillions" of games.

BTW I learned how to play to Go, but I suck at Reversi and even Connect 4.:(

Don't Drick and Drive.

Please Log in to join the conversation.

More
4 years 7 months ago - 4 years 7 months ago #9833 by Phoenix Spiritus
Phoenix Spiritus replied the topic: What we now know about Magic... Sorta...
Actually, all that is still "calculating", the computer is getting input, applying a programmed amount of calculations to it, and then delivering an output.

If you took the Google go computer, and replaced the Go board with a new board, say Monopoly, without a human reprogramming that computer it can't do anything. All the same calculations that let it win Go can be used to let it win Monopoly, but it takes a human to repurpose them and let the computer start accepting Monopoly boards as the input and giving Monopoly moves as the outputs.

This is what I mean by computers can't "think", they only calculate. If someone hasn't programmed them to do something with a particular input, they can't do anything with that particular input. At all. Ever. A human needs to come along and reprogram them to accept it.

The "intelligence" of a program is purely a reflection of how well programmed it is. How many possible scenarios the programmers thought of and wrote code to handle. If you can be more inventive then the programmers, your will find ways to give input to any computer it hasn't been programmed to handle. And no matter how many times you give that same input to them, without a programmer updating the program, the computer will stay unable to handle it.

Proper "thinking" machines will be able to change their reactions through repeated stimuli to eventually learn how to handle the new input.
Last Edit: 4 years 7 months ago by Phoenix Spiritus.

Please Log in to join the conversation.

Moderators: WhateleyAdminKristin DarkenE. E. NalleyelrodwNagrijMageOhkiAstrodragonNeoMagusWarrenMorpheusWasamonsleethrOtherEricBek D CorbinMaLAguASouffle GirlPhoenix SpiritusStarwolfDanZillaKatie_LynMaggie FinsonDrBenderJGBladedancerRenae_Whateley
Time to create page: 0.119 seconds
Powered by Kunena Forum