It's 2005, why does the AI still need to cheat?

[Zoid]

-Ab-, No, but people need to get in touch with reality. And if they want to moan they first got to code a better AI than the one that´s in Civ4...

[MxM]

Originally posted by rockhopper
Isn't Civ4 in its entirety itself a "cheat?" I mean, you are not really a world leader, are you?

You mean I am not? Darn! And I though why they laugh at me when I ask them to build a palace for me!

[Sleestax]

I must have done a poor job of expressing my idea because people seem to be drawing the exact opposite conclusion from my proposal.

It is exactly because it is hard that I am proposing that Firaxis open the AI code to the open source community. Maybe they can pre-qualify a set of programmers, project leaders, etc... It can only help them in the end. It would cost them little to manage and they could ignore the result if they didn't like it.

However, open to a group of very dedicated programmers who don't have to worry about minor details like a budget for salaries, I think the AI could be massively improved.

It's not as if I am proposing something new. Linux has been operating this way for years.

Constructive replies are prefered, but I've been around long enough to expect mostly flames and/or derail.

[Sava]

it's 2005 and we still use our hands to play video games?


THATS LIKE A BABY'S TOY

[Gaal Dornik]

Exponential problems cant be solved with simple raw computing power. This just aint gonna happen any time soon. And for such complex game like Civ4 most probably never. huge maps, lots of units, lots of moves. Forget it.

[Kuciwalker]

Re: It's 2005, why does the AI still need to cheat?

Originally posted by Sleestax
I understood, in the bad old days, why the AI needed to cheat. Computing power was low and heuristics were a budding concept.

What I don't understand is why the AI must "cheat" or to be nicer, be given certain advantages, given the state of technology today. Most gamers PCs are as powerful as a supercomputer was 15 years ago. Memory is abundant, gfx cards are powerful, etc...

That's frees up a lot of CPU for a smarter AI that doesn't need hardcoded "cheats" built into it.

I understand that this is a business and not an experiment. Perhaps Firaxis would open just the bit of code for the AI to the open-source community to allow them to build a better AI for Firaxis for free.

Opinions?

You're an idiot who knows nothing about AI is my opinion.

Next.

[GodSpawn]

It's not as if I am proposing something new. Linux has been operating this way for years.

I don't believe that part-time programmers would come up with a better AI than Civ's.

It they could, wouldn't they already have done so with FreeCiv? I think most people play FreeCiv in multiplayer - as far as I know, the AI is not up to Civ IV standards.

[GodSpawn]

[dearmad]

Without cheating by looking into the current solutions used. Try this: program a path finding algorithm that has as much time and processing as you want to give it.

I've done this. I came up with a pretty good one that could solve any maze of obstacles with a perfectly efficient path that took into account many different influences (terrain cost, preference to be direct or indirect, avoid certain "dangers", efficiency, etc.). In other words, it wouldn't "ride the wall" and go into concave spaces and resort to retrograde motion in order to get anywhere thorugh any path.

It took me about three months to do this.

THREE MONTHS for a freaking path finding algorithm!

Now, what sort of heuristic is involved in the code that would allow the computer opponent (CO) to select (using a relatively flawless pathfinding algorithm), WHICH of the various successful paths it should take and why? When it has roughly 100 paths... even 10 paths to choose from, which is preferable and why? Go directly at the city? Go around to the back via the land? Take a boat? WHICH boat? The nearest one nearer the enemy front? Or the boat parked out back behind the friendly capital city? Etc., you get the idea.

THIS is one way to begin to understand how difficult an AI is to program when you haven't designed the game to be played by the CO (such as games like Corporate Machine are).

And I've only touched on the surface of how to integrate a pathfinding algy with the CO's choices of HOW to use it.

Speed and memory ARE helpful in solving these problems, but the real weakness is in the human being's understanding of efficacy in playing games- efficacy: efficiency, effect, and effectiveness are balanced in proportion while game playing in ways that no math has yet been able to define. One of the fundamental truths to games is that, for example, efficiency isn't always best when deployed against superior effect. Another exzample is that poor effectiveness (a "ploy" in other words) can maximize effect or efficiency elsewhere. *YOU* do the math on that, if you can. I bet you can't- nobel prize winners can't.

Even the economist/physicists who DO win nobel prizes on this front have only very BASIC mathematical models- such as the ploy of risk versus reward- that are anywhere near to being functional, applicable solutions. And they assume you know what RISK is versus REWARD... which for a sodomist meeting a mosochist (for example), as you can imagine, isn't totally up front.

[nbarclay]

I posted the following in another thread regarding AI "cheating":

Developing a great AI is a whole lot easier said than done. In chess, each turn, each player makes one move. In Civ, each turn, the player can make up to tens or even on larger maps hundreds of moves (including not just units but build orders, changes in civics, and so forth). In chess, you can see exactly what the situation is at any given time. In Civ, unless the AI cheats, there are a lot of things about what the other players are doing that the AI doesn't know. In chess, there are a maximum of two players. In Civ IV, there are up to 18. In chess, there are 64 squares. In Civ, there can be thousands or even tens of thousands. In chess, the exact outcome of every move is known in advance. In Civ, some types of moves have a random element involved in their outcomes. And so on.

When you compare the games' complexity, Civ is enough more complex than chess to make Deep Blue (the computer that could beat the world's top chess players) look slower than an Apple II. Using anything resembling a brute-force approach, a Civ-type AI cannot possibly consider more than a tiny fraction of all the different combinations of things that could happen over the next several turns.

That means AIs in Civ-type games are very heavily dependent on programmers to tell them what strategies to use in different types of situations. But even if the people who provided the strategy for the programmers were the best Civ players who would ever exist, AIs would still face several problems. First, something would inevitably get lost in the translation between what goes on in the best players' thinking and what gets programmed into the AIs. Second, the AIs would still be stuck working from general rules rather than being able to invent new situation-specific strategies on the fly. Third, in the absence of a genuinely massive investment in giving the AIs a variety of alternative strategies, human players could take advantage of learning to recognize patterns in the behavior programmed into the AIs. And in any case, after the game's been out a while, there will be human players who are better at the game than anyone was at the time the AI was originally developed and who share their knowledge with others.

So with anything resembling current artificial intelligence technology, it's just plain unrealistic to expect AIs to be able to compete with the best human players with both sides playing by the same rules. We can expect AIs to come closer over time as computers get more powerful and AI programming techniques improve, but if top players want AIs that can give them a serious fight, we have no choice but to give the AIs advantages on the higher difficulty levels.

[Harovan]

Originally posted by Fried-Psitalon
Lastly - although this debate is certainly ongoing - I would argue that in fact the AI does not cheat in CIV.

It seems to see through the fog of war. I have a savegame proving this. This counts as rule cheat (not bonus) in my books.

[kimmygibler]

The ai WILL be open eventually will it not? Then mabye we will see improvements.

[Gaal Dornik]

Imagine you have 1 simple warrior. Each warrior can move to 8+1 fields. For the AI to see whats going on next turn there are 9 possible positions to calculate. No probs. Now lets take 2 warriors 9*9=81. With 3 warriors thats 9*9*9=729 possibilities.

Now lets take a normal map, 7 nations, warrior+settler. A settler is faster, allowing him to go to 25 fields. makes it (9*25)^7=29'192'926'025'390'625= 29 Peta-possibilities. Todays PCs can do like 3gHz( 0,000003petahetz) And that excluding everything else(like graphics, music..)

Making todays PCs twice as fast, like 6gHz will not even scratch the above number.

Well, that was for one turn at the beginning, imagine late game with 100+units, railroads, and deep 10 turns AI strategic. Thats a number highly unimaginable.

[Bhruic]

You think programming an AI is easy?

Then let's perform an expert. Can you beat the AI on 'Noble' (the level closest to no AI-cheats)? If so, do you ever lose? If not, do you cheat in any fashion (reloading, etc)? If not, explain exactly how you would, if you were the AI, and the AI you, have stopped you from winning. Now assume that the AI did that, could you have stopped it? If so, come up with another strategy that the AI could use to stop you from stopping its original strategy. Figure out how the AI would know to use either strategy.

And I could keep going.

The reality is, even the best AI is limited in what it knows to do. If you look closely at the majority of 'strategy' that gets presented for Civ IV, you'll note that most of it involves how to exploit the system. The AI, however, doesn't have the ability to learn how to exploit the system. Nor does it have the ability to figure long term strategy.

Personally, I find the AI to be competent enough in Civ IV. I know it's going to press me, and I can control how much it does so by adjusting the bonuses (or penalties) it gets. And I lose the odd game, which keeps me playing, becase I know I'm going to be challenged.

I'm not sure what more you could ask for in an AI.

Bh

[-Ab-]

/me climbs into flame proof suit!
ok if you got 2 warriors and the first one has 9 choices,why then does the second multiply it by 9?
why is it not 9 choices and then 9 more choices?
3 warriors have 729 possibilitys? give an example of that

and whos peter ?