Printable View
I am playing a campaign with a friend right now and just about every battle where the AI attacks with a general, they send the general ahead, discover the enemy army, turn around but instead of going back, they start running in circles, getting their general unit shot up by archers and by the time the rest of the infantry arrive, the general is dead.
It's rather difficult to program AI to handle battles well because the battlefield is so fluid. If you consider what AI has challenges with, it's almost always modelling that entails a lot of variables, weather, economics, just too much going on. Given a more controlled setting, say a chessboard, it will mop the floor with you. :yes:
Its not that bad. While the battlefield AI isnt exactly stellar it will at least replace the general's scouting role with another cavalry unit as soon as its available and when faced with enemy matchlock cores my experience is that the general hides behind his lines until the teppo are engaged or can only get 1 shot off if he charges. Sure, he might die from that one shot... but its better than constipating on your toilet and having the world think a ninja stabbed you in the arse from your toilet drain.
The AI is smarter than we think...
Yeah because for more controlled settings, the AI can "think" ahead to the point where everything you do is met with the perfect counter. On this game, it feels like the AI isn't even "trying". They just follow the same pattern for just about every battle, wether my army consists of 2000 ashigaru or 1000 samurai with cavalary support, spears, archers and matchlocks or wether they are sieging or attacking me on the open field.Quote:
Originally Posted by easytarget
To make the game harder, the AI isn't made smarter, it is just given bonuses and your unit's stats are reduced. I dunno since when companies make their games 'harder' that way but I find it extremely lame and annoying.
True enough, it's a common problem in game design, in fact there was another thread here someone started discussing (as I recall) games that had effective harder settings.
The solution, and this is me speaking from no knowledge of how this works, would perhaps be to play back into an AI's strength by pre-setting tactics based on watching hundreds or even thousands of human vs human matches (not like CA couldn't collect the data). The patterns of engagement aren't infinite, or even close to it, that would then be the tactics you would program to, making it a series of steps the AI follows. And of course it would be even better if they could figure out a way to tie back results to modifications in subsequent battles the AI fought, course I think that would be rather hard. So the brute force approach would be the first place to start.
The problem is that game designers/publishers are generic liars on this subject. They call their progams and algorithms 'Artificial Intelligence' when there is no intelligence in them. They do not learn, they cannot 'think' strategically. They only do what their pre-programmed number crunching dictates (including the latitude of randomness they are programmed with). What everyone in the gaming industry calls AI is NOT. Don't perpetuate the myth.
The problem is that intelligence and thought processes are difficult concepts to define let alone attempt to translate into programming. Calling game designers "liars" is rather harsh. If you contest their use of the term "artificial intelligence", what would you propose they call it?Quote:
Originally Posted by feelotraveller
Thought I'd clarify here =p I don't mean the entire AI army... a lot of the stuff they do "makes sense". I am talking about the general unit in particular, which they seem to treat as a glorified unit of light cavalary.
TW has long had a problem with suicidal generals. It was one of the, relatively few, problems with the original STW AI. If you were unlucky, you could end up fighting an entirely ronin Japan, the factions proper sacrificed their generals so frequently. However, I confess I did not notice the problem with STW2. It seems to use the general unit, as in ETW and NTW, for behind the lines support rather than as a combat unit.
If the AI general is just exposing himself to your fire, you could always refrain from shooting at him. (A British artillery captain at Waterloo observed Napoleon parading in front of his troops across the ridge and asked Wellington if he could take a shot at the enemy general; Wellington was aghast and refused permission[1]). It's more of an issue if - as in the original STW - the general charges into your lines.
[1]By contrast, in the ancient period, trying to take out the enemy general and decapitate their army was a common battle tactic.
It's helpful to remember that history is replete with foolish, stupid generals. Few were Caesars. Custer at the Little Big Horn borrowed a page right out of an AI battle plan. So did the French at Agincourt. And who could forget Cannae? One could go on and on.
So in this regard, the often AI does a good job simulating commanders!
I like to win my battles when I play. I don't mind an occassional gaff made by the AI generals. I have made a few myself actually, but no Napoleon am I!
Yes intelligence is hard to define - self awareness, creativity, learning... there are many possible characterisations I agree. It is harsh to call them liars but accurate. Take an example from S2TW - if I was playing my 4 year old nephew, it might take him a while, but on the 10th siege (or maybe more) he would resist lining his archers up to face my general and ignoring the main body of my army. (Lucky I don't play him at platform games, he would whip my b***.) The 'computer program' (my suggestion because that is what it is) will never learn this and, outside of reprogramming intervention, will fall for the same mistake endlessly. An artificial intelligence worth its name would function 'as if' it was human. It would 'learn' = reprogram itself. Sadly the game designers know this and wish to perpetuate the myth that you are somehow playing something intelligent rather than an inflexible coding.Quote:
Originally Posted by Gregoshi
The sad thing is that eventually we all start to think that the computer program is really an artificial intelligence because it is repeated endlessly. Yes it is a complex and remarkable piece of code but in the end it is no different from any unthinking mechanical contraption. Would you want to call a mouse trap intelligent?
Feelotraveller,
Not to put too fine a point on it, but you do realize you're not exactly making an earth shattering revelation here right? You act incensed as if you were somehow duped into buying a game with big bold letters on the box falsely proclaiming the game inside has human like AI.
The next thing you'll reveal in full outrage will no doubt be the epiphany that most games feature character and plot development so infantile in comparison to other more developed mediums like literature or cinema that we should all gather round and have ourselves an ole style game bonfire and torch the lot.
But before we consider anything so drastic, here are some possible alternatives to belaboring your point in this thread any further:
1. check the box in S2 that says drop in (next thing you know a human like me will do just that, and drop in to replace the AI)
2. play chess, the "AI" will be more than happy to curb stomp you till the cows come home
3. or maybe just stick to MP style games in general, thereby avoiding the dumb AI altogether
Just a reminder, let's keep the discussion to the issue, not each other.
To add a couple of more TW items to easytarget's list:
- Don't exploit known AI issues.
- Attack! Attack! Attack! Perhaps it is a false impression on my part, but most accounts of dumb AI actions or general battle descriptions seem to be with the player sitting in a nice comfy and defensible position and letting the AI take the initiative. Then the player just takes advantage of any poor execution on the part of the AI. Take the initiative yourself and give yourself more of a challenge. Let the AI react to you or see how well you can react to changing situation rather than just standing in a perfect defensive position. Put yourself in a position to potentially make a boneheaded move. Get out of your comfort zone.
As for the AI learning, I agree with Nelson's point that history is filled with incompetent commanders, so the AI need not be brilliant in most cases. Add to that that dead commanders can't learn (unlike your 4 year old nephew), assuming the AI would be simulating each individual commander rather than a vitual player such a we are in the game (controlling all battles).
Ah... use the AI's 'stupidity' as part of the immersion.. I hadn't looked at it that way... Thus far I was all about MUSTWIN!!11 =p Probably my own mistake o\
Ah, Easytarget and Gregoshi you miss my point. If I tweak the difficulty levels S2TW can give me more than a run for my money. I am not unhappy with the game. (In terms of challenge it beats my nephew hands down.) I generally do avoid exploits, etc.
It's just not an AI I am playing. It wouldn't be if I was playing chess or any computer game that I know of.
The definition of AI is to be human like (or to pass as human -be indistinguishable from one- in certain circumstances) with regards to intelligence. One way we could understand this is if 'drop in' were turned on and I couldn't tell whether I was playing a human or not. (In which case it would not matter, challenge wise at least, whether it was turned on or not.)
No, it is not an AI, because I imagine, that even if it is possible to make an AI (which would then become capable of doing pretty much anything), you needed very strong machines (far beyond any normal person's budget I assume) to 'make it work'.
Video games simulate an AI, which works pretty nicely up to the point where you know all the of its tricks.
Is it even possible to beat a computer at chess? I mean they simulate any way a chess game can go from the current set of pieces, so they always make the best possible move. The only limit is their own speed I guess and that is increassing by the day.