Chibi-Matoran wrote:

King_Nether wrote:

DaSpudLord wrote:

King_Nether wrote:

WooHooBoy wrote:

...

Great question, it depends on the situation, if it's gonna lose or die by moving forward, it will move backwards.

And for the chess thing, that was an example. It will try to win the game of chess. It could be smart, but not smart enough to conquer Russia or USA.

Remember playing a game against a computer, thats pretty much what the CPU block does. It's supposed to be a Computer player.

DaSpudLord wrote:

Yes, but how would it make that decision? What kind of factors would weigh in on its decision to move either forwards or backwards? How would this information be delivered to the AI? We're asking you to think of the actual implementation and usage of the block, but you're too focused on actually getting it to figure out how it would actually be used.

^^^
Building on to what I said, what about knowing what's the enemy and what's not the enemy? How woud the AI be able to know what's good and what's bad? How does it know who it's enemies and who it's allies are? The AI would have to be programmed extremely well to be able to do this without being given any special instructions, so… how would it get those instructions? How would it know what it can and cannot do? How would it know what powers/powerups it has? There is no way that the ST could make an AI smart enough to figure out all of that on its own.

What about an avoid sensor and a go to sensor. Basically, if the avoid sensor is in the script, it will avoid it, if there is a go to sensor, it will try to go to it.

That wouldn't work well for a Chess AI, because different pieces have different moves; a knight can't simply move around the screen to avoid a queen. Sorry, a one-size-fits-all AI block isn't very feasible.

Also, you could probably program such a block yourself with ease.

Well, what about an option, board game like or video game like.

King_Nether wrote:

Chibi-Matoran wrote:

King_Nether wrote:

WooHooBoy wrote:

King_Nether wrote:

Chibi-Matoran wrote:

Sorry, but your idea is too broad to be made into a single block. A Chess-playing AI is probably programmed completely differently than a Goomba or an enemy spaceship. An entire project can be devoted to an AI.

For example, an AI program in another programming language can potentially be made up of many different programs working together, composed of many classes extending or implementing each other and containing members of types from even more classes, as well as so, so many functions calling each other. Considering one function is equivalent to one block, an AI program is too huge for merely a block.

Even a simple AI which can be written as a single function would be hard to implement as a block which can be used widely by many Scratchers. Let's say I program a Chess-playing block. What kinds of arguments would it accept? How would the board be represented? With a list? How about the various pieces? Are they clones, or separate sprites? How the Chess data is represented may vary between Scratchers based on their individual coding styles. Designing such a block which can be easily used by all Scratchers is not feasible.

Also, CPU means Central Processing Unit, and is a part of a computer.

Okay, thank you. But consider the following, if the AI had to choose to move forward or backwards, it would choose the best choice for itself.

But what _defines_ the _best choice_? Is its best choice to move forward every time? Is it to move backward? How would it be able to handle a game like chess, where there are so many combinations you can't be sure?

Great question, it depends on the situation, if it's gonna lose or die by moving forward, it will move backwards.

And for the chess thing, that was an example. It will try to win the game of chess. It could be smart, but not smart enough to conquer Russia or USA.

Remember playing a game against a computer, thats pretty much what the CPU block does. It's supposed to be a Computer player.

Sorry, but problems still exist. What is “win?” What is “lose?” These conditions depend on the project, and the project genres which can use an AI is too broad for an exact definition. In Chess, winning is checkm8ing the opposing king and losing is getting checkm8d. In addition, those motivations can be divided into even smaller tasks; the AI must strategize to capture other pieces, and avoid getting captured. What if would it want its own piece to be captured on purpose? In a spaceship game, the AI would be completely different! Remember that computers do not have common sense; they take instructions literally and seemingly obvious steps must be defined at the lowest level.

The example about win and lose was not chess. Win I talked about was the exit, lose is dying. So the AI will try to not lose, but infact try to win. If take what I said to a different matter, you may understand what I am talking about.

That is the problem I was trying to explain to you. An AI for one game will not work for another game.

King_Nether wrote:

Well, what about an option, board game like or video game like.

Too many games exist. What about Mancala? Can the block play checkers? The point of Scratch is to learn how to program, and adding a block for every single need hurts this purpose.

Chibi-Matoran wrote:

King_Nether wrote:

Well, what about an option, board game like or video game like.

Too many games exist. What about Mancala? Can the block play checkers? The point of Scratch is to learn how to program, and adding a block for every single need hurts this purpose.

Well, you can program the AI, kinda like a modified broadcast.

Well, please add on the fixes of these problems.

King_Nether wrote:

Chibi-Matoran wrote:

King_Nether wrote:

Well, what about an option, board game like or video game like.

Too many games exist. What about Mancala? Can the block play checkers? The point of Scratch is to learn how to program, and adding a block for every single need hurts this purpose.

Well, you can program the AI, kinda like a modified broadcast.

Well, please add on the fixes of these problems.

You can program the AI yourself.

Chibi-Matoran wrote:

King_Nether wrote:

Chibi-Matoran wrote:

King_Nether wrote:

Well, what about an option, board game like or video game like.

Too many games exist. What about Mancala? Can the block play checkers? The point of Scratch is to learn how to program, and adding a block for every single need hurts this purpose.

Well, you can program the AI, kinda like a modified broadcast.

Well, please add on the fixes of these problems.

You can program the AI yourself.

^^^
Listen, there is no way to make a perfect AI that would cover every kind of project, or even a majority of projects. There are too many contingencies possible for a one-size-fits-all AI. Sorry- it's not getting implemented.

This is a really good suggestion. Unfortunately, I see some problems.

From what I am seeing, there is too much ambiguity in your defining. Every single project will need a different code for each AI to work properly.

Take Siri for example. It uses a set of programming AI made specifically to assist a user.
Then, move to Mariokart- If I used the “Siri Block” to program the game, the functions would “explode”, and the game would not work properly.

So sadly, No support.

However, I would like it if there was some sort of guide to make AI's. Both simple, and complicated. That way, people can learn how to make an AI system without it being done for them. It might be a bit hard a first, but they would get better in the end.

Chibi-Matoran wrote:

Sorry, but your idea is too broad to be made into a single block. A Chess-playing AI is probably programmed completely differently than a Goomba or an enemy spaceship. An entire project can be devoted to an AI.

For example, an AI program in another programming language can potentially be made up of many different programs working together, composed of many classes extending or implementing each other and containing members of types from even more classes, as well as so, so many functions calling each other. Considering one function is equivalent to one block, an AI program is too huge for merely a block.

Even a simple AI which can be written as a single function would be hard to implement as a block which can be used widely by many Scratchers. Let's say I program a Chess-playing block. What kinds of arguments would it accept? How would the board be represented? With a list? How about the various pieces? Are they clones, or separate sprites? How the Chess data is represented may vary between Scratchers based on their individual coding styles. Designing such a block which can be easily used by all Scratchers is not feasible.

Also, CPU means Central Processing Unit, and is a part of a computer.

^^^^ As per above

So your basically suggesting, to have the scratch team create a sentient intelligent AI that can read people's minds.
Even though this would be nice to have, it isn't possible with today's technology.

No support

No way.
-Is this really worth the ST holding up everything else and spending a bajillion years on this?
-How does this sense and react to specific things?
-If you're thinking this big, can't you just do it yourself?

Zro716 wrote:

you're basically asking the ST to create a fully conscious and intelligent program that can build other intelligent members.

please tell me you're joking

scrooge200 wrote:

DaSpudLord wrote:

How would this work? How would the AI know what to do and what the rules of the game are? How would you define these things? How would you define the abilities of the AI? Would you be able to program special power-ups or other special occurrences in the game? The problem is that each game is different and unique in its own way, and therefore, each game needs its own special AI programmed so that it works with that game. Sorry. No support.

Don't you have anything to say besides what's already been posted?

I see many posts here explaining why this won't work, so I feel I don't need to explain either. I will just say that, as a programmer of an AI for a game, this idea simply won't work.

And yes, I will use the argument that if you want an AI, you should learn how to program it yourself. In other words, if they could add some sort of AI block, they could also have an “animate” block or “simulate” block.

no support. why?

Too much ST coding
it could be used for millions of things, making millions of different AIS,
Also, every game is different. you can make a small ai that does
pick random number and then for each number have a separate output

No support. Each game requires a specific AI. The ST can't just guess what kind of game you're going to make and add tons of AIs for that.

No support as per everybody.

liam48D wrote:

No support as per everybody.

Every other ATer hates you for summing it up

Absolutley no support, this is VERY hard to accomplish on quite a lot of programming languages, and flash/python is no exception.
In fact, I'm not even sure it is fully possible on python, but maybe on flash.

If it was possible, I predict it would cause an outcry in the community. People seeking the so-called “fame” within scratch would be absolutley astonished that their hard work to seek their “fame” is dwarfed by people who could master the AI block, this also implies that the AI block would need a lot of user programming to work properly. It would cause chaos to the entire website, with those who mastered the AI block will have their projects showered in love-its and favorites from new scratchers looking for inspiration, while those who have not, flood the forums with various threads by older scratchers protesting against this. If not removed, the scratch forums will eventually just become some kind of abandoned section of the scratch website, and eventually it will move to the projects, and then finally, the wiki, and after that? well…? Who knows what might happen?
Here are a list of kind of “Endings” to the AI block catastrophe

1. The website will be completely abandoned by normal scratchers, essentially meaning survival of the fittest, and the diversity of the scratch website languishes, therefore meaning the diversity of the projects start again from square one, Games and Art, the other more diverse projects become relics of the past.

2. The AI block is removed, therefore turning scratch back to normal, but with something to remember better than that flamewar…

3. Everyone Masters the AI block eventually, while this would confuse new scratchers, the community faces very little change from before the AI block was added.

Despite the last and second ending, there would be a lot to repair, so no support

SuperNicky wrote:

Absolutley no support

If you're gonna no support, then there's no reason to necropost.

SuperNicky wrote:

If it was possible, I predict it would cause an outcry in the community. People seeking the so-called “fame” within scratch would be absolutley astonished that their hard work to seek their “fame” is dwarfed by people who could master the AI block, this also implies that the AI block would need a lot of user programming to work properly. It would cause chaos to the entire website, with those who mastered the AI block will have their projects showered in love-its and favorites from new scratchers looking for inspiration, while those who have not, flood the forums with various threads by older scratchers protesting against this. If not removed, the scratch forums will eventually just become some kind of abandoned section of the scratch website, and eventually it will move to the projects, and then finally, the wiki, and after that? well…? Who knows what might happen?
Here are a list of kind of “Endings” to the AI block catastrophe

1. The website will be completely abandoned by normal scratchers, essentially meaning survival of the fittest, and the diversity of the scratch website languishes, therefore meaning the diversity of the projects start again from square one, Games and Art, the other more diverse projects become relics of the past.

Dang, talk about excessive… You're basically talking about a Scratchpocalypse. While there might be some upset, I doubt it will be that bad…

You'll have to make your own AI, because making an AI block would do some crazy stuff, and would take an infinite time to perfect. When humanity is able to make sentient AI, then, MAYBE, you'll get an AI block. It'll only be, oh, you know, 18623513475162783564 years!

not all games are alike, how would the scratch team make a “cpu” block for every single type of game if there are uninvented types of games