Opinion: Why should I care that the AI doesn't cheat?

I'm not really hung up on whether a game AI cheats or not, but definitely prefer if it doesn't. There's a few reasons why I and I suspect most people would care.
1) Even though this is a single-player game, in my mind the AI is a substitute for playing against a real player. You can see a cheating AI as equivalent to doing handicapping, but I think many people would prefer to be challenged without the need for handicapping. Whether it's a strategy game, or an actual sport like golf, or bowling, I'd rather play straight against someone of my own skill level.
2) The need for an AI to cheat usually indicates that it is deficient in some ways. The cheating might enable it to still be a challenge, but this still unbalances the gameplay in that it can train you or even force you to exploit the AI's (sometimes obvious) weaknesses rather than play what might be the optimal strategies against an equal opponent.
3) #2 above often means that the player's options become more restricted against a cheating AI: Sometimes perfectly valid tactics against a human or non-cheating AI become infeasible and unworkable against a cheating one, and sometimes expert players learn to avoid certain tactics (or call them "cheese") if they want a challenge, because the cheating AI is unable to handle them at all. Here's a really basic example: Let's say a game has units with some concealment feature that gives them limited undetectability, but the cheating AI always sees the whole map and all units. Those units lose their big advantage and probably are no longer worth using, if you're playing against the chetaing AI, even though they might be great against a human or a non-cheating AI.
4) When the AI isn't playing by the same rules I am, it breaks some of my immersion in the game, particularly if it's blatant/obvious. It strikes me as: "What's the point of playing against the AI, if it is playing by different rules than I am"?

Well said Orion Adrian, absolutely correct and I completely agree. I like cheating AI, bring it on, its more challenging and more fun.

There is something more satisfying about being challenged by an AI that cannot use any tactics unavailable to you. What is especially impressive is when you encounter an AI that can defeat you without using any tactics that are unavailable to human players.

good points Lintman. I totally agree with point #3 and #4. If the AI cheats by getting for money or resource per turn or at the start more than me, I don't mind that. But if the AI sees everything, then what's the point of giving us invisibility units? They immediately become things that are just for show. Another cheating thing that I really hate is the one in CIV3 (not sure if 4 still got it) that the AI nations have preferential treatments with one another. It totally doesn't make sense that they are trading techs when they fought several wars just recently. But they won't trade with me even though I am their best pal till now.

LintMan says it all.

handicapping and cheating are not the same. handicapping is giving some players advantages, like golf score or more resource per turn off the bat, or taking your rook out of the chess game. cheating is when some players play with different rules than others.

Pretty cool argumentations!!!

I'd add another quote, which is: If the AI is not cheating, everything it does, every step it takes, every evolution rate it achieves can be also achieved by the player, provided that random events don't unbalance things drastically. I've been playing GalCIV (I) for a while and got astonished at several facts, such as the rate with which the AI was able to put way more colony ships and constructors in orbit than I was capable of in the first turns. That made me reconsider my tactics. Just because I was assured the AI was not cheating. And it was even playing on Normal difficulty!.

So in my experience, every game can be an inspiring one on itself, because it makes gamers, at least those not quite veteran such as me, reconsider our tactics and improve our overall strategy just by taking a look at what the AI is capable of doing without cheating.

I think that in order for a computer to stand a chance, it must cheat. It should get extra bonuses. Although it shouldn't trade with itself and exclude the human like in Civ III, it should at least charge the human more per trade. In GalCiv, I notice that some of the AI's are pathetically weak. They are so far behind in tech, that they just become super tempting targets. I can't help but play evil because it is so easy. And it definately needs a production bonus because it is so inept that if it were producing at my level it wouldn't even put up the pathetic fight it does with a production bonus. I think this game is very difficult for the AI. I won't say the AI is bad. In fact, it is most likely very good. However, the game is too complicated for it. Untill AI is truly "I", Give it all the advantages it can have I say.

And for now, I trust that the devs are doing everything that can to improve on the AI. This game is fun even if I win too easy. I can't wait for the next patch.

LintMan is spot on IMO.

I've always hated AIs that cheat outrageously, but only because I hate players that cheat against me as well. Yes, I know the AIs in games tend to be stupid and need some advantage against good players, but allowing them to do things that are impossible for the player really rubs me the wrong way. To me, its the same as people teleport hacking in WoW, using aimbots in a FPS or using a fog of war cheat in a RTS. Its a huge, unfair advantage that to me ruins the enjoyment of the game.

On the other hand, in this game you can consider the AI to be "cheating" at high levels in one respect: it gets more money. At the highest level, signifigantly more money. This doesn't rub me the wrong way because its a single factor, easily understood, that doesn't interupt the game or change the basic building blocks of the game. I'm already used to racial bonuses and resources that do simmilar things, they just get a permanent one as an advantage. Its not hugely unbalancing, but it helps compensate for the fact that I can think better than the AI and can guess at the AIs moves (of course, I might be able to say the same about a human as well). Its more of a handicap in the way you'd use it in golf or in sports games, instead of the way you'd use it after losing three limbs. Something to even the playing field, without asking me to play the game with one hand tied behind my back hopping on one leg. Sure the game is hard either way, but its hard in the "I need to be better at the game" sense, not the "I need to learn how to hop really well" sense.

If game developers start taking the easy way out by making the AI cheat instead of giving it intelligence will end up slowing down the creation of an AI that actually is truly intelligent simply because fewer people would be working on it and giving it such a high priority. As it is the AI's are getting smarter than gameplay is complicating but if you say cheating is fine intelligence doesn't matter developers could and many would for efficiency just make AI's complicated enough to use the game mechanics then make them cheat because they can't master the mechanics.

Not speaking about any specific game in particular, but if the only way a game can increase its difficulty level is by creating an inequitable shift of the rules in favor of the computer, then that is not what we should call "artificial intelligence". Perhaps we should adopt the term "artificial opponent"; that would be more accurate than describing a cheating enemy as "intelligent". Again, still NOT speaking about any game in particular.

So Stardock is responsible for SkyNet!

For the sake of the human race, stop making the GalCiv AI smarter!

I hate it to play against a cheating AI. I never play on difficulty levels where I know the AI will be cheating. And actually I avoid games at all if I know the AI is so badly implemented that the AI has to cheat even on normal levels.
For me it is extremely important to have an AI which play by the same rules. I would waste my time playing against a cheating AI and what does it prove ? For me it was the info about the good AI in GalCiv2 which made me buy the game because all to often I see lame AI in other games where they try to compensate it with mulitplayer instead. I don´t play mulitplayer. I want good, turn-based single player games with good AI. I have been brought up with old good strategy games since the Commodore 64/Amiga era where we had a lot of them from SSI, SSG and others (Second Front, Western Front etc from Gary Grigsby)

The AI cheating can put the advantage too far in the AI's favour. This doesn't worry me in the level of challenge, but it would very much frustrate me if, say, the AI built something in the space of one turn that would normally have taken two. When the rules go out of the window, the player has to learn not only how to play as a human, but how the computer can play as well, in order to make informed choices during the game. Personally, this does not strike me as being synonomous with the idea of a strategy game.

There is a big, BIG difference between playing a game of basketball with a klutz who is a foot taller than you, or an opponent of your own size and skill level.

Playing against an opponent who plays by the same rules is immeasurably more immersive because it tells so much more of your own abilities and decisions. If he outsmarts you, then you chastize yourself for not seeing it coming, and you learn to try it yourself the next time. If you outsmart him, you realize that you just defeated someone who had all your strengths (and weaknesses). It's more immersive because you are, on some level, playing a mirror image of yourself, and you learn a heckuva lot about your own play in victory and defeat. You don't feel the same impact when the cpu wipes you out at the beginning simply because they started out with three times the resources.

Before the rebuttals start up, I realize that you can still have a challenging, interactive, learning experience on an unlevel playing field, and if you're really good, maybe that's better. But I hope you get my point.

I think that about solidifies the arguement.

I also agree LintMan is spot on. A reason why good AI is best for arcade style fighting games, which is very very rare.

Anyone played Age of Empires on hardest against the AI?
It builds the heighest tech units about 30 seconds into the game whilst you've still got guys with pebbles as your best units. It teleports troops into the back of your base. It moves every one of its units simultaneously And you can still beat it.. It is a challenge, but there are only a limited range of options available, so there is little replay value... you can only watch the computer send wave after wave of his units at your massive defence peacemeal before getting bored.

Now compare with the AI in warcraft III for example. It doesn't cheat, and it is as good as (or better than) many newer human players. It feels very similar to playing a human.. which is what we'd all really like to be doing.

A good AI is really about variety in tactics - it should adapt to what the human does so that it doesn't always fall for the same old tricks, and can continue to provide a challenge as you get better at the game.

Bad AI = repeatedly falls for the same tricks/exploits (no challenge).
Good AI = variety (challenge).

When it comes to the AI 'cheating' I would say it is not too important, as long as it isn't too obvious. However, a tough AI on an even field is surely the ultimate target for any self respecting AI developer

Cheating is bad if it lets the ai do/avoid things you cannot do/avoid. I want to play by the same rules as the ai, I don't want to play chess against an opponent that plays go. I have no problem if the ai starts with more/less cash/units, in better locations, or has production bonuses/maluses to provide for some challenge.
As for the original post "Fairness doesn't apply to a computer" I disagree. Computer games are programmed by humans, and playing against an ai is playing against the human team that programmed the ai. They are at a disadvantage because they have to think of all their moves beforehand, but they are still human and there is no excuse for them to break the rules.

So Stardock is responsible for SkyNet! For the sake of the human race, stop making the GalCiv AI smarter!

It's ok to make automated factories, and it's ok to make killer robots, but it's not ok to make automated factories that make killer robots.

A cheating AI limits my options as a player. If the AI knows where my units are, I can't hide things or set up ambushes. If the AI gets free money every turn instead of having to tax/trade/mine/whatever for it, I can't choke their economy to slow down their war machine. If the AI always knows what I'm doing, deceptive strategies become impossible. An AI that doesn't cheat can be fought using many different strategies. An AI that cheats is only beaten by exploiting some stupidity on the part of the AI, which makes for a much less satisfying victory.

For me it is simple: I can learn from an AI that beats me fairly, but I can't learn from an AI that beats me by cheating.

A lot of good points in the posts upto now. I especially like to emphasize that while I don't care about give the AI a handicap when it needs it (like giving it the economic bonuses at higher difficulties), but it would really suck if the AI can call in reinforcements from another galaxy and have fleets coming out of nowhere.

That said, perhaps the most innovative part of the GalCiv2 AI is that it is scaled downwards, from intelligent to just stupid. Yet, how many new players have you seen coming in asking how the heck did the AI beat them senseless when playing on normal or lower? Now, assuming that they didn't go asking for tips and kept trying, I assume they will be able to learn how to play the game just from watching how the AI does it. See, when an AI that plays fairly beats you silly, it just means you're doing something wrong, and needs to play better. But if an AI that cheats beats you silly, that just means you should lower the amount of advantage you give it at the start (The Civ AI anyone?), learn its weaknesses on those difficulties, then crank the difficulty back up and exploit what you've learned.

The OP is right in that, in the end, a single player game is about the experience, and a challenge is fun. However, it is twice as fun to be schooled by a really good AI that just plays better than you... I'm not saying the GalCiv2 AIs are perfect, but it has provided some really fun moments while learning how to play the game. Aside from covering up its weaknesses which are being addressed in the patches, the only thing I would change is the ability to choose the AI and it's economic advantage/disadvantage separately. That is, it might be fun to play a game with an intelligent AI who's running at 100%, and a fool AI running at 200%, just to see who's tougher. I can see a lot of possibilities with an option like this.

Errrrrrrrrrrrrrrrrrrrrr, look just play the game to overcome. I don't enjoy the winning so much as the struggle to win. What ever they have to do to beat me I will figure a way out of it, or I will get beat in rare occassions. Life is a struggle, the game is set up that way. It is the best turn based strategy game I am played. Enough variables and customization to allow anyone to overcome any obstacles.




"Long live the fighters!"

Muadib, Hous Atriedas

From an academic perspective, it is of relatively little interest compared to the impact AI has on the gaming experince, there game developers not research graduates. Gal Civ II's AI wont be the centre of discussion anywhere else than in the games industry as that is where its only real practical application lies.

Lets look at this from another angle, your playing a racing game, you and the computer have exactly the same car, same modifications even the same paint job, yet his car can go 50mph faster than yours. So by default its taken all the strategy, quick timing, skill out of a driving games AI, and made up for it by putting this 50mph speed boost in the AI's favour to make it challenging.

A cheating AI is a poor substitute for a well thought out strategic, skillfull, manipulative, suprising porting human-like behaviour and tactics. And people wonder why online games are so popular these days compared to single player games. Because in most cases the next best thing to playing a human just isnt up to the challenge, however in Gal Civ II's case it takes a big step forward.

In Addition to the other good points made by Empyrean, Kalin, tekito, kankerfist and Lintman's arguemtn which ive regurgitated in a few instances by coincidence.

P.S while i personally dont care if you personally dont appreciate this, i care about what others will gather from reading about why a good AI (or set of different AI's) is an essential part of any single player game.