Fugitive Sisyphus
Escape Artist
That is not true. Everything that can be done on a quantum computer is already possible on a your basic personal computer.
Civ 5 is not even Civ anymore.
- Thread starter Bradicus
- Start date
That is not true. Everything that can be done on a quantum computer is already possible on a your basic personal computer.
Hail
Satan's minion
ai will always remain stupid and the only way to compensate for ai's stupidity is bonuses. ai development is long, complicated, and the result may "misbehave". in short: ai does not make sales. the only good thing devs can do, is release the civ's ai interface (and the std ai dll code
), so that anyone can code an ai dll and attach to the game engine: much like how it is done in c-evo.i bet after a while, some enthusiasts will pop up and improve the ai on their spare time.
this idea fits nicely with "user-driven content" paradigm. strange devs do not that already.
maybe they do not want to release some ingenious algorithms in the ai dll deanej
Deity
Nope. For example:
The only reversable operator on a current computer is negation. On a quantum computer, all operators are reversable except measurement.
However, the reverse is true: anything that can be done on a current computer is possible on a quantum computer.
Great, go build one then
In the mean time, I'll re-iterate that quantum computers are NOT required for advances in the development of more advanced AI than that found in Civ IV.What you're saying is comparable to the arguement "I need a more fuel efficient car" with your reply being "Dream on! Hydron fuel cell powered cars are just too advanced and not worth it yet!"
You're creating a strawman argument. You're not identifying anything to do with the topic of this thread, instead you're creating you're own argument that subtly shifts focus from the point. Sure you are correct when you say a very advanced AI could be created with a quantum computer. But lets be realistic. Is it REALLY required for a solid AI to be created? Hell no! If American programmers are too lazy to do it, I'm sure firaxis could outsource the job to India for a third of the price! It is doable. Saying something is so difficult or hard to do doesn't make it not worth doing
deanej
Deity
Actually, we're a lot closer to quantum computers than most people think. I estimate within 10 years we might have the first consumer quantum processor. Unfortunately the magazine article I was going to link to isn't online yet.
In either case, my point still stands: how would you go about coding a better AI? I don't know of any better way to make an AI than what is presently in civ. We can only try to make a better AI on the system we have now; a better system will have to wait.
In either case, my point still stands: how would you go about coding a better AI? I don't know of any better way to make an AI than what is presently in civ. We can only try to make a better AI on the system we have now; a better system will have to wait.
Sorry if I sounded snippy, I wrote a philosophical logic final today and I'm glad that chapter of my existence is over for good!
Alright, I profess my relative "noobness" in the realm of coding and computer science; that being said, are their not other innovative ways in which to "build" on the AI of the current system? Create more complex coding to amplify the intelligence of the current AI? I was under the influence that just about anything can be programmed with enough time, coffee, and beer.
What if a rival company to Firaxis wanted to create a civ-like AI but better? Would they only be able to match the current AI of Civ IV? Or would they only get something as good? It can't be bested with current technology? I'm actually curious about this now. *shakes fist*
Alright, I profess my relative "noobness" in the realm of coding and computer science; that being said, are their not other innovative ways in which to "build" on the AI of the current system? Create more complex coding to amplify the intelligence of the current AI? I was under the influence that just about anything can be programmed with enough time, coffee, and beer.
What if a rival company to Firaxis wanted to create a civ-like AI but better? Would they only be able to match the current AI of Civ IV? Or would they only get something as good? It can't be bested with current technology? I'm actually curious about this now. *shakes fist*
deanej
Deity
I think what your suggesting basically what the BetterAI team is doing now. IMO it's an enhancement of the current AI rather than something completely different.
Fugitive Sisyphus
Escape Artist
Operators on a quantum computer would certainly not all be reversible unless you mean by reversible that the computer holds the belief states representing all the valid inverses of an operation but in that case a classical computer can do this too, just less efficiently.
If you don't believe me wikipedia says "Quantum computers however do not allow one to compute functions that are not theoretically computable by classical computers, i.e. they do not alter the Church-Turing thesis. The gain is only in efficiency."
Hail
Satan's minion
the time and effort required to code and modify an algorithm is proportional to the exponent of it's complexity
that is why the KIS (Keep It Simple) paradigm is so important
GOAP beats hierarchical SM's any day
putting time and effort in eye-candy will pay off more, than putting the same time and effort int ai. more so, the devs are not interested money-wise to make games with infinite replayability.
Perhaps it would be possible to develop a better AI by having human players teach it. Here's the idea: make some kind of AI scripting interface for the next Civ and a game mode in which players can send their AI's against each other in an MP game. What you do is make an AI duriing single player games that copies the algorithm's you use to play, then pit your AI's against each other.
The interface could be like, as you play a single player game under the "AI development" custom option you can't make moves directly, you have to create an algorithm for the move to be based on. For example, you can't just build a temple, you have to create a rule mandating that if you have an unhappy face in your city your next build must be a building that makes happy faces, and then you must let the now amended ai execute that rule. Later, you may have to amend that to, say, have an exception for circumstances of war.
The interface could be like, as you play a single player game under the "AI development" custom option you can't make moves directly, you have to create an algorithm for the move to be based on. For example, you can't just build a temple, you have to create a rule mandating that if you have an unhappy face in your city your next build must be a building that makes happy faces, and then you must let the now amended ai execute that rule. Later, you may have to amend that to, say, have an exception for circumstances of war.
deanej
Deity
Then wikipedia is wrong, as it contradicts the sources I used to make a presentation on the ethical implications of quantum computing last October. That statement isn't cited, either, which leads me to believe that someone misinterpreted something they read when writing the article. If you have a subscription to Wiley InterScience, please see: http://www3.interscience.wiley.com/cgi-bin/fulltext/122328239/HTMLSTART
Fugitive Sisyphus
Escape Artist
I can't see the article.
I wish you could explain more fully the example of all operations being reversible on a quantum computer. Going by how I interpreted what you stated a quantum computer could for instance reverse logical AND to get the inputs 'true' and 'false' solely from the result 'false'. This should be impossible. What is possible is that you can deduce that the inputs were 'true' and 'false', 'false' and 'true', or 'false' and 'false' and if more information is found later you can further reduce the set of possible inputs. Logic systems can and have been made on classical computers to solve this type of problem.
What I think this is about is that there are some problems(many to do with artificial intelligence) which are known to have no polynomial time(P) solutions. What this means is that the time it takes to solve the problem increases exponentially with the input size. This means that even though a classical computer can solve such a problem you are not going to see it implemented in an AI because it would take too long. However a quantum computer might be able to take advantage of nondeterminism to solve some of these problems in polynomial time.
deanej
Deity
I don't really know how it works, but it probably has to do with quantum bits being "fuzzy" (not either 0 or 1, but every value in between all at the same time). This is also apparently the reason for them being faster.
Schroedinger's wave equation includes a square root of negative one, meaning there is a value that could be either -1 or 1. Practical quantum computers will supposedly take advantage of this to do more than one thing at a time (how many depending exponentially on the number of operating qubits). Some say this is because the computer is working in league with alternate universe copies of itself. It doesn't matter how it works under the hood, though. Its sufficient that the potential is there. However, I have read articles not too long ago about breakthrough new quantum computers that had just a handful of qubits. Practical multipurpose quantum computer's aren't something we can expect soon. But I don't think that precludes good AI, which is a software problem. While there are quantum processes involved in and influencing the brain, they usually average out. For all intents and purposes, most people use algorithms and heuristics that can be replicated by software. Qualitatively, AI approximating average human intelligence is a software problem. Some creative and problem solving human thought processes involved comparing lots of factors simultaneously, which capability emerges from how the brain works, but this is because it uses spreading activation not because it uses qubits. Conventional computers must execute tasks in serial order, but this can be used to simulate many simultaneous comparisons because individual comparisons can be made so quickly. Very large problems or inefficient processes are slow on software that runs on common personal computers, but there is no need for a perfectly stimulating game AI to be unable to avoid such pitfalls.
One of the problems with Civ is the sheer size of the map. A first person shooter is doing lots of graphics, but AI wise there may be a handfull of bad guys making decisions about which rock to hide behind or something. All that is is what you see, plus enough behind it to make it plausible. In Civ, the entire iceberg is really there under the water, in all its glory. In civ you have maps with dozens of civs making decisions about hundreds of cities with thousands of units and tiles, and decisions should be strategic, each taking into account lots of other factors. Better BTS AI is noticably slower than the standard AI and all it does is force the AI to think about things a little bit more. In this case, a quantum computer really would be better, because the AI is simulating more than one person thinking deeply about lots of complicated situations, and trying to do it instantly. So what's the solution? How could a better Civ AI be done with software? What about multiple AIs? There could be one AI for anything that will be seen by the human player, and another for the submerged part of the iceberg. Two AI civs fighting a war on the opposite side of the world from the human need not deeply ponder strategy. All that matters to the human is who won and how it affects the human.
One of the problems with Civ is the sheer size of the map. A first person shooter is doing lots of graphics, but AI wise there may be a handfull of bad guys making decisions about which rock to hide behind or something. All that is is what you see, plus enough behind it to make it plausible. In Civ, the entire iceberg is really there under the water, in all its glory. In civ you have maps with dozens of civs making decisions about hundreds of cities with thousands of units and tiles, and decisions should be strategic, each taking into account lots of other factors. Better BTS AI is noticably slower than the standard AI and all it does is force the AI to think about things a little bit more. In this case, a quantum computer really would be better, because the AI is simulating more than one person thinking deeply about lots of complicated situations, and trying to do it instantly. So what's the solution? How could a better Civ AI be done with software? What about multiple AIs? There could be one AI for anything that will be seen by the human player, and another for the submerged part of the iceberg. Two AI civs fighting a war on the opposite side of the world from the human need not deeply ponder strategy. All that matters to the human is who won and how it affects the human.
trickofthehand
Prince
- Joined
- Oct 16, 2007
- Messages
- 310
Why wouldn't the Devs do this? If they expect me to do it they can pay me a salary as well, and some points on the back end.
Eh, yeah, hopefully. It just seems to not make any sense for them not to make Civ V, given the huge, ginormous market potential of the game.
Lockesdonkey
Liberal Jihadist
I like how I'm the only one who's made the argument on actual "Civ ought to remain Civ" grounds rather than babbling about some technology that your average fellow has never heard of (I've heard of it, but I have computer nerds for friends and uncles).
Gamemaster77
PC > Mac
I am actually expecting to be disapointed when it comes out so I am not expecting something great and don't get it. If only they could get all the developers from civ1-4 to make civ5...That would produce some nice results if they stay away from alot of civ4 gameplay and use thier features instead.
Similar threads
