1.1 - AI still not challenging :(

According to the wiki, no, the AI won't get any smarter as you turn up the difficulty past tough. But it will get more powerful, which will give you more of a challenge.

You have the advantage of having a brain developed over millions of years of evolution, capable of learning, pattern recognition and abstract reasoning. We're a long way from getting that into a computer.

To compensate for your evolution enhanced brain, handicap yourself by playing at a level above tough. The computer will still make some dumb moves, but will offer more of a challenge.

Yep, just what justkevin has said.

On higher difficulties the AI might become bonuses (lots of them on suicidal), but it really makes it tougher and better. The AI also plays somehow smarter then. Not because of better algorithms, but because it just has more things to use and throw at you if nescessary.

In my last game for example the AI declared war against me in the early game and just a turn later it's fleets (7 ships in each fleet armed to the teeth, ie. 270 beam damage fleet) arrived at my border worlds with escorted troop transports right behind it. I was going for a research victory this time so hadn't researched much into the weapons tree and built even less ships. I still was ahead in weapons research tough, so I've just built some big bad dreadnoughts of doom. Well, not only didn't I had the time to build enough of them (the AI took three planets in the second turn after declaring war) I didn't even had the nescessary logistic points to destroy its fleets. My ships may have been twice as strong as his, but I was unable to build larger fleets then three of them so when I was attacking I still lost some ships from his counterattacks... and sence he had about a gazillions of ships I couldn't protect my ressource bases... well, the rest is history of an embarassing defeat.

Needless to say that the AI didn't wan't to make peace, no mather what I was offering him...

But I just hurt him at last, because I've offered my surrender to the second powerful empire, which with my planets hopefully revenged me.

----------

The game is still enjoyable on those higher difficulties because the AI only recieves bonuses, it doesn't cheat. What's the difference you might ask. Well, it doesn't receive ships out of the blue, it doesn't have unlimited money, etc. The AI still plays with the same rules as you, it just has some quite high boni in the economic, miniaturisation, sensor range and building (not sure about the building thing) part of the game. And because of that it's still beatable. (Even though certain victories other than military victories may become a bit harder if you not pay attention.)

This is not an excuse.
Computer are perfectly capable of learning, identify patterns and think abstractly, and also guess, try, feel frustrated, scared, bored, angry. It's just need to be programmed that way.
We are not a long way from that. But generally, who have knowledge do not have the money to implement that. Who have the money is too restrained by the budget constraints.

Sorry, it's the truth.

Kisses from me...

Edna

PS: Capitalism Sucks... Technocracy is the Key

This is not an excuse. Computer are perfectly capable of learning, identify patterns and think abstractly, and also guess, try, feel frustrated, scared, bored, angry. It's just need to be programmed that way. We are not a long way from that. But generally, who have knowledge do not have the money to implement that. Who have the money is too restrained by the budget constraints.

Ah... yea. Sorry, must have missed my last courses about AIs.


I don't want to sound insulting... but is it possible that you have read a bit too much sci-fi novels? (Not that I wouldn't do this of course. I'm just curious.)

Beside AI. This game suffers from a serious balancing issue.
The opponents are just too strong or too weak.
Wins who can advance more quickly in the Research Tree. Seens obvious. If one manages to rush the Tech tree without been treatened by military opposition, is a serious candidate to invincibility.
It's hard to play cooperatively because the stronger opponents are never willing to assist the weaker and this goes against what happens in reality.
In real International Relations, the Balance of Power theory states (among other things) that the second stronger powers will gang against the the strongest one and the strongest will gang with the weakest in a symbiosis relation.

I am getting away from the point.

In my opinion two colonies should not add to one another but to integrate one another.
For example, two research labs each producing 10 units should not result in 20 research units but in 15. This could keep the expansionists from becoming invincible and would make the small nations more important. And I think it something more realistic because if you have two labs in real life, they will not research twice faster, but a little faster because they will exchange ideas.

Thanks for listening,

Kisses from me...

Edna

don't want to sound insulting... but is it possible that you have read a bit too much sci-fi novels? (Not that I wouldn't do this of course. I'm just curious.)

You don't sound insulting. Your comment and curiosity are both acceptable.

I am a software developer for 12 years now and game addicted for the same 12 years. I back my posts with a lot practical knowledge and successfuly tested software algorithms.

Your interest are very well accepted.

Kisses from me...

Edna

Edna,

I do not know what you do in real life, but communicate your ideas is obviously not one of them.

So, read what your worte in the last two posts, and then throw it away. Your thought are intelligible.




I love sci-fi, I truelly do, but if we were as close to your Ideal AI, or maybe you were close to said AI.. you would have been a government/military employee a very long time ago.


Not to insult or anything.



-N

Well in my first game in 1.1, at the "tough" difficulty, the iconians made fleets... Lots of them. They made fleets with 8 ships each, combining big fighters and small fighters, and they actually did a good job fighting my surprise invasion... They even attacked my home world while i was attacking their planets. I think the AI improved a lot, and the game is really really fun as it is. Sure, it is perfectible, but face it, this is the best AI that exists in video games as of today.

lol...

maybe you are right... should not post while working

Kisses anyway

In real International Relations, the Balance of Power theory states (among other things) that the second stronger powers will gang against the the strongest one and the strongest will gang with the weakest in a symbiosis relation.

Yes, this does happen. But it makes for a pretty boring game. With equally matched alliances forming (as would happen in real life --- think pre WWI Europe) there would be long periods of peace since a smart AI would know that it could never gain a strong advantage over another alliance. It would make diplomacy/relations, which is hard to duplicate in a game, much more important.

I'm willing to accept a little less reality to make for a more enjoyable game.

Hello Haasen,

You got the point.

there would be long periods of peace since a smart AI would know that it could never gain a strong advantage over another alliance

In my opinion this would be nice because that would give time for other nations to develop social, military and research. But as you said, this could become boring.

An on/off option would be great.

Kisses from me.

Edna the Intelligible.

May be there is still an AI bug ?

See: v1.1 Difficulty/Computer Intellegence Level Bug? (https://forums.galciv2.com/?ForumID=274&AID=115007)

It looks like the AI is not initialized in some games ...

Thanks for the replies guys.
To everyone who told us about your personal positive experience with the AI - I'm happy for you, but when I play, the AI really doesn't do anything smart.

According to the wiki, no, the AI won't get any smarter as you turn up the difficulty past tough. But it will get more powerful, which will give you more of a challenge. You have the advantage of having a brain developed over millions of years of evolution, capable of learning, pattern recognition and abstract reasoning. We're a long way from getting that into a computer. To compensate for your evolution enhanced brain, handicap yourself by playing at a level above tough. The computer will still make some dumb moves, but will offer more of a challenge.

I do not see what this has to do with evolution. The AI does not need to learn stuff in-game. The game rules are predefined and the amount of distinct tactical situations not infinite. So yes, given good coding I don't see why AI can't be better than human at GCII.

On higher difficulties the AI might become bonuses (lots of them on suicidal), but it really makes it tougher and better. The AI also plays somehow smarter then. Not because of better algorithms, but because it just has more things to use and throw at you if nescessary.

I'll try playing at higher difficulty, but somehow I don't see how the braindead AI will become better because of that.

May be there is still an AI bug ? See: v1.1 Difficulty/Computer Intellegence Level Bug? (https://forums.galciv2.com/?ForumID=274&AID=115007 Link) It looks like the AI is not initialized in some games ...

I doubt that's the case, since the AI tried to do something, but did it in a half-assed way. The Thalans whom I didn't attack did build around 5 fleets and had them hanging in space around their planets.

You could stop playing Yor for one thing.
Being able to build Colony Ships with 8 movement by turn 4 doesn't make the game harder exactly.

In my own experience, the AI is much superior to 1.0 although I still have no problem beating the crap out of them one at the time. Haven't played any game far enough to see if all the remaining AI:s will unite against me. That will probably make it a little harder to win by military conquest.

Edna hasnt been reading too much sci-fi, YOU have not read ENUGH sci-fi!
What you would know then is the amazing fact that the real world and the sci-fi line is a very very fine constantly blurred line. We have the capability for a self-evolving AI, we just dont have the funding for it.

Its just like when people said that you couldnt go to the moon, or tought that the stars were a flat sheet over your heads.. You will be amazed at what can be acomplished in the next 10 years. Did you seriously believe we would have graphics and physics in pc games like we do now, 10 years ago??
10 years ago we sat and played 2d platform games, and anything more than 10 sprites on screen lagged
There was NO physics other than a static gravity. And there was no interaction at all between actors, it was just "flameball.gif hit mario.gif, delete flameball.gif delete mario.gif, load mariodead.gif"

I do not see what this has to do with evolution. The AI does not need to learn stuff in-game. The game rules are predefined and the amount of distinct tactical situations not infinite. So yes, given good coding I don't see why AI can't be better than human at GCII.

While this may be true, your computer's CPU can only proccess so many of them each turn. If the AI acutally looked at all these opions at end of each turn, you'd be waiting for 5 minutes each time.

I think that FrogBoy said in a recent journal entry that he had got them to add some new options to allow more more CPU time devoted to the AI in the next update. Maybe that will give them the advantage you want.

lol... maybe you are right... should not post while working Kisses anyway

No doubt

My experience.

V1.0 -

Game 1 Played at 'normal' with humans customised as research race got trashed. When the AI came down on me like a hawk while i was trying to turtle.

Game 2 Played at 'easy', with custom race, gave up midway when i found myself overwhelmed.

Thought to myself, the AI is really really good. Got humiliated, despite being a veteran of turned based games since the original civilization, master of magic, master of orion etc. I won all of them at the highest difficulty level though you might say with cheesy tactics, but to be beaten by a AI not even at full strength....

Game 3 Swallowed my pride, Played at 'cake walk', started colony rush strategy ,won without problems.

V1.1

Game 4 played at 'normal' colony rush with 'economy' race. Won easily. The key is to make sure you are not last in the military standings after you have colonized the prime colonies. But i read they toned down the normal and bright AI in 1.1.....

Game 5 played at 'challenging' setting using same strategy. See above.

Game 6 played at 'tough' setting (3 opponents at intelligent) again using same strategy. Same result. , apparently high diplomacy skills can hold them off even with no/weak military.
And no, I didn't do anything 'cheesy' either to win (unless getting god like diplomacy is considered cheesy). I barely did any tech trades despite way better diplomacy skills.

I'm wondering if it will work at higher levels, i suspect it might just not as easily.

Verdict : The human brain is still better at spotting weaknesses.Once I got used to the unusual economy system, got familiar with the tech tree, figured out the 'trigger points' in which AI declares war and play around that, it is not that hard to win. And from what i read on the forum, with players finding 'sucidial' easy, it seems the AI has even much bigger weakness that's i'm not exploiting such, as speed + scanners for instance ,,,,,,

Still, the AI here is certainly the best of any game, I've played barring Chess AI. But I think some people expect too much, if you are a TBS player worth the name, after a couple of games, no AI should be able to equal you on even terms.

I must admit, I too had unrealistic expectations, particularly after how some reviewers raved about the AI, and after my first 2 unsuccessful games, but it was merely me adjusting to a new economic system , once i got to grips with it....

I'm always curious to know exactly what in the world people were expecting from the AI when they make these sorts of posts. It's real easy for the human player to make the sorts of long ranging tactical decisions that for the most part in a game AI that has to run in a matter of milliseconds are impossible. Also keep in mind this AI is NOT scripted in any way. Brad has made numerous posts as to why so I'm not going to bother repeating them.

I don't know about the rest of you but I don't have a HAL9000 sitting on my desk. From my perspective since it runs on my machine without taking a day to figure out what to do I'd say this AI is if not the best one of the best ever written for a strategy game.

Now could they cheat and make it more of a challenge? Well of course. But that would defeat the purpose. Also it is a game. And paramount to all else a game must be fun. An unbeatable AI would not be fun.

To Edna:

Computer are perfectly capable of learning, identify patterns and think abstractly, and also guess, try, feel frustrated, scared, bored, angry. It's just need to be programmed that way. We are not a long way from that. But generally, who have knowledge do not have the money to implement that. Who have the money is too restrained by the budget constraints.

Would be really curious to know what computer runs this sort of thing. If it takes it more then a few milliseconds to *think abstractly* then all I can say is not in my game . And the last time I checked unless you've got some top secret project knowledge while it is possible for a computer to be programmed to mimic one or two out of you list there has been no computer to do them all. So yeah I'd say we are still decades from this from a computing power standpoint. Not to mention the fact that ok yeah it can feel fear but can it also walk up a flight of stairs? Isolated instances of a particular algorithm aren't what I'd consider viable for making a game *build better ships*.

I'm always curious to know exactly what in the world people were expecting from the AI when they make these sorts of posts. It's real easy for the human player to make the sorts of long ranging tactical decisions that for the most part in a game AI that has to run in a matter of milliseconds are impossible.

Dev quote from https://www.galciv2.com/screenshots/gameplay/PowerDuel.html

I set the Drengin to tough. This is their best game without getting freebie money and such. Only around 5% of users should be able to beat it at this level.

This is what I expected. Somehow I don't think the 5% number holds.

Also, the AI doesn't have to make long ranged tactical decisions in milliseconds. Since usually there's no hurry to make those, the AI has plentyof time to think during my turn (and I believe it does already).

Also keep in mind this AI is NOT scripted in any way. Brad has made numerous posts as to why so I'm not going to bother repeating them.

You're mistaken. If it wasn't scripted in any way, it would be a self-learning neural net, which it is clearly not.
While it's not scripted, as in "do this then that then that", its possibilites are scripted.
For example, the AI has scripts telling it to put ships in fleets, and scripts telling it to send transports to the enemy planets. The AI's ability is defined by the quality of its scripting, and it can never do anything outside of it's scripts.

Now, what would a truly non-scripted AI be?
An non-scripted AI would probably only given the rules and objectives of the game, and develop its own ideas of how to play it.
A non-scripted AI would be able to understand and counter any human strategy.
A non-scripten AI would be able to use any human tactic including the cheesy ones.
Such an AI is impossible to make for GC2 of course.

Now could they cheat and make it more of a challenge? Well of course. But that would defeat the purpose. Also it is a game. And paramount to all else a game must be fun. An unbeatable AI would not be fun.

A cheating unbeatable AI would not be fun. However I'd love to see a fair-playing AI beat me at GC2.

Just play on Masochistic. It's plenty hard enough.

While this may be true, your computer's CPU can only proccess so many of them each turn. If the AI acutally looked at all these opions at end of each turn, you'd be waiting for 5 minutes each time. I think that FrogBoy said in a recent journal...

Would be really curious to know what computer runs this sort of thing...

Today I written a quick and inteligible post (while working) disagreeeeeeeing (do this word exist?) of the bullfrog about AI and CPU Times, please refer to it for some comments:

Link

You would be amazed to see the wonders that the 80386 computers were capable of doing.

Today, with multi-threading, the computer can "think" while we are thinking. What I mean is that the computer can gather information while the player is Idle. And with "Idle" I am including the time between mouse clicks.

How long take a player's turn? Don't you think it is enough for the computer AI to elaborate an strategy on their own? And if the CPU don't have enough time it is not a problem. If the computer have to spend one o two turn to think of their strategy don't you think this is acceptable for the computer to take wrong decisions? This sort of thing can make AI feels even more like humans.

It is possible. We done it already.

but if we were as close to your Ideal AI, or maybe you were close to said AI.. you would have been a government/military employee a very long time ago

To reach a position like that one, being close to this ideal AI is not enough. One need money to know important people or have the luck to be relative to one.

Kisses from me...

Edna

As Haasen stated in a previous post in this thread, with a Balance of Power logic "there would be long periods of peace since a smart AI would know that it could never gain a strong advantage over another alliance".

The true challenge both for the player and to the CPU would be to get the upperhand and to obtain supremacy over all other races. This would be extremely dificult and In my opinion would be the extase of playing a game like this.

If you don't understand the logic, think about it:
In a game with 10 races, gang the most powerful civilizations against all others. If they are more powerful, keep balancing with the weakest civilizations until the to sides are equals. When the most powerful civ is no more then rebalance and do it again.

Kisses from me...

Edna

I set the Drengin to tough. This is their best game without getting freebie money and such. Only around 5% of users should be able to beat it at this level. This is what I expected. Somehow I don't think the 5% number holds.

Well if it said, 5% of users playing their first or maybe second game, i would say maybe. But like any AI, that is not adaptable with time, once you play enough you start spotting it's weaknesses, for example, most players I think after a couple of games get a sense on the conditions it will declare war, and work to make sure it never happens etc. T

If you know generally how your opponent is going to react, it's a HUGE advantage.

Personally I think i was a bit scared off by the reputation of Galciv's AI in my first 2 games, so I gave up too early. Plus in the first game, I didn't even figure out how to invade colonies until much later.

In fact, it's highly beatable even from an inferior (not vastly of course) position, but I'm seldom in that position for TBS games. I I've replayed some of my saved games, and it's beatable with a quick transport blitz.

You're mistaken. If it wasn't scripted in any way, it would be a self-learning neural net, which it is clearly not. While it's not scripted, as in "do this then that then that", its possibilites are scripted. For example, the AI has scripts telling it to put ships in fleets, and scripts telling it to send transports to the enemy planets.

Probably a semantic matter. By not scripted, most people mean that it does not do x,y,z actions regardless of what is happening, but rather it takes into account a lot of factors in its decision.

In that sense, it's much like a human being, like when choosing to research technology, we take into account the overall position, whether we are at war, our overall victory plan ,whether we can or should get it from trading etc.

I suppose these are embodied in APIs like CalcFullAttack(), CalcFullDefense().

I doubt if Galciv2 is the first game AI to do this of course, but properly done it can lead to a respectable opponent.

You are right in that the AI cannot from game to game, add its own APIs, while humans can figure out new strategies, take into account new factors from game to game, but that is not a bar from high performance, since top Chess AIs can't do this either.

Now, what would a truly non-scripted AI be? An non-scripted AI would probably only given the rules and objectives of the game, and develop its own ideas of how to play it. A non-scripted AI would be able to understand and counter any human strategy. A non-scripten AI would be able to use any human tactic including the cheesy ones. Such an AI is impossible to make for GC2 of course.

A human isn't scripted (by your definition)? I would argue that using a wide definition of 'scripts', even neural nets are fundamentally, 'scripts' at the lowest levels, just doing their tasks stupidly...

The nice thing to have of course is some kind of persistent self learning program, that can figure out it's errors, and adapt to the human. Chess AIs generally don't do this , they have positional learning but this applies only to a specific position, which is less useful since it can't generalize.

Experiments in neural nets and other forms of machining learning for chess AI have generally not being as successful as standard alpha-beta programs with hand tuned Evaluation functions. That's basically the same as GalcivII really.

You could turn on TT since the AI was programmed with that in mind (turn it off makes the AI much weaker in my experience). You could also try playing with a non-Yor race since their mini-bonus makes them pretty much the strongest. Even so, the AI does some things a bit wrong (not putting enough emphasis on diplo techs for example)