Slightly OT: AI Rant
Moderators: Joel Billings, elmo3, Sabre21
-
- Posts: 6948
- Joined: Thu Jan 17, 2002 10:00 am
- Location: Cottesmore, Rutland
Slightly OT: AI Rant
When I first got into computers I did so because finding a board game opponent was as rare as rocking horse defecation. I read an article, think it was old S&T mag, in the early 80 about computer gaming and thought that looks like what I need so bought myself an Apple ][e complete with a few 5¼ floppy discs. This is it I thought the future in gaming.
Two games from the mid 80, Gary Grigsby’s Kampfgruppe and War in Russia spring to mind immediately. Both classics and considering the size of the computer memory then (128K in the 2e) the AI played reasonably, they had too as there was no email or internet. Yes there was such a time. Not all computer game were good then some real dogs where the AI was almost no existent.
As memory got bigger games got more complicated costing considerable more to develop. Games became prettier (seems to be the main criteria for a successful game these days), certainly play well against another human player, most lack fatal errors that made some games unplayable and they are all definite more stable. What they do lack is anything resembling an even poor AI certainly nothing that matches an old game’s AI, but I’ve got a very leaky memory or should I say selective memory according to my wife.
When are designers, programmers et al going to remember all those people who do not want to play another human but the AI. Spend some money on developing a decent AI please and not those scripted ones, but something that can make a decision based on the tactical situation.
Two games from the mid 80, Gary Grigsby’s Kampfgruppe and War in Russia spring to mind immediately. Both classics and considering the size of the computer memory then (128K in the 2e) the AI played reasonably, they had too as there was no email or internet. Yes there was such a time. Not all computer game were good then some real dogs where the AI was almost no existent.
As memory got bigger games got more complicated costing considerable more to develop. Games became prettier (seems to be the main criteria for a successful game these days), certainly play well against another human player, most lack fatal errors that made some games unplayable and they are all definite more stable. What they do lack is anything resembling an even poor AI certainly nothing that matches an old game’s AI, but I’ve got a very leaky memory or should I say selective memory according to my wife.
When are designers, programmers et al going to remember all those people who do not want to play another human but the AI. Spend some money on developing a decent AI please and not those scripted ones, but something that can make a decision based on the tactical situation.
RE: Slightly OT: AI Rant
Yah.
I have to chuckle at all of these futurists like Ray Kurzweil who keep trumpeting what all of these superintelligent AI's will be capable of. Meanhoo in our wargames AI hasn't really undergone any sort of quantum leap of any sort at all in 2+ decades.
Yes, a Grigsby wargame is going to be harder to code than chess or go is, and would basically necessitate budgeting the same amount of time, manpower, and $$$ as the core game. What they really need is a generic modular learning AI that can be plugged into almost any game and can progress (vs. being utterly stagnant and thus eternally predictable).
I have to chuckle at all of these futurists like Ray Kurzweil who keep trumpeting what all of these superintelligent AI's will be capable of. Meanhoo in our wargames AI hasn't really undergone any sort of quantum leap of any sort at all in 2+ decades.
Yes, a Grigsby wargame is going to be harder to code than chess or go is, and would basically necessitate budgeting the same amount of time, manpower, and $$$ as the core game. What they really need is a generic modular learning AI that can be plugged into almost any game and can progress (vs. being utterly stagnant and thus eternally predictable).
RE: Slightly OT: AI Rant
Unfortunately, I don't think you should hold your breath for a Machine Learning (ML) based AI in WitE.
First, the technical side. In order to train a ML bases AI, you would need quite a lot of training material. In games such as Chess, Go and Texas Hold'em (poker) such material is readily available and all of those games have seen a ML based AI developed. So for a wargame to have a ML based AI you would need to release the game, have a lot of people play it (while collecting data) and then set the ML based AI to train. Next, one of the absolute strengths of a game such as WitE is that the rules are continously changing. For an example, the HQBU's rules have changed significantly over the last year or so. So an ML based AI trained on the old rules would make some pretty silly mistakes in a game with the new rules. Another reason why it is not so simple to just do an "intelligent" game AI. Finally, I am wondering if the strategies used in a game such as WitE are simply not to advanced for anyone to train an AI quite yet. With a success criteria of being in Berlin by 1945, the Soviet ML based AI would be swamped with options to such a degree that training it would be virtually impossible.
Second, the business side. As ML development is still in it's infancy, there are simply just a lot better business opportunities to throw ML resources at before anyone would get around to games such as WitE. Next, the training would be a problem from a business perspective as well. Obviously, you can't release a "blank" AI. For the first many games, such an AI would start the game as the Axis by moving half it's units in the direction of Berlin. Not really sound from a business perspective, as noone would buy such a game. In short, the game would still need a scripted AI when released. The challenge would be to build a ML based AI on top of the scripted version in such a way that it could slowly replace the scripted version. That would then reveal the next problem. If such an ML based AI was finally trained, noone would be able to beat it. Wite is a game won by the side that makes the least mistakes. A ML based AI would slowly become better to such a point that it could never be beat (provided that training material etc is there).
As for writing a generic ML based AI that could play any game, I just can't see it happening. It would be like writing a piece of computer code that could do whatever task you set it. You would still have to code in victory conditions and "AI allowed moves". But even so, the internal algorithms would be general and not necessarily suited to the specific game. I just don't see it happening.
In short, I would guess it will still be a decade or so before games such as WitE will se a functioning ML based AI. However, the development of ML is going so fast that I might be surprised.
First, the technical side. In order to train a ML bases AI, you would need quite a lot of training material. In games such as Chess, Go and Texas Hold'em (poker) such material is readily available and all of those games have seen a ML based AI developed. So for a wargame to have a ML based AI you would need to release the game, have a lot of people play it (while collecting data) and then set the ML based AI to train. Next, one of the absolute strengths of a game such as WitE is that the rules are continously changing. For an example, the HQBU's rules have changed significantly over the last year or so. So an ML based AI trained on the old rules would make some pretty silly mistakes in a game with the new rules. Another reason why it is not so simple to just do an "intelligent" game AI. Finally, I am wondering if the strategies used in a game such as WitE are simply not to advanced for anyone to train an AI quite yet. With a success criteria of being in Berlin by 1945, the Soviet ML based AI would be swamped with options to such a degree that training it would be virtually impossible.
Second, the business side. As ML development is still in it's infancy, there are simply just a lot better business opportunities to throw ML resources at before anyone would get around to games such as WitE. Next, the training would be a problem from a business perspective as well. Obviously, you can't release a "blank" AI. For the first many games, such an AI would start the game as the Axis by moving half it's units in the direction of Berlin. Not really sound from a business perspective, as noone would buy such a game. In short, the game would still need a scripted AI when released. The challenge would be to build a ML based AI on top of the scripted version in such a way that it could slowly replace the scripted version. That would then reveal the next problem. If such an ML based AI was finally trained, noone would be able to beat it. Wite is a game won by the side that makes the least mistakes. A ML based AI would slowly become better to such a point that it could never be beat (provided that training material etc is there).
As for writing a generic ML based AI that could play any game, I just can't see it happening. It would be like writing a piece of computer code that could do whatever task you set it. You would still have to code in victory conditions and "AI allowed moves". But even so, the internal algorithms would be general and not necessarily suited to the specific game. I just don't see it happening.
In short, I would guess it will still be a decade or so before games such as WitE will se a functioning ML based AI. However, the development of ML is going so fast that I might be surprised.
To be is to do -- Socrates
To do is to be -- Jean-Paul Sartre
Do be do be do -- Frank Sinatra
To do is to be -- Jean-Paul Sartre
Do be do be do -- Frank Sinatra
RE: Slightly OT: AI Rant
Old games were more simple than wite. That makes making a good ai easier. In general terms the ai is extremely bad at attacking, and bad while defending.
The basic problem for having a decent game against an axis ai is that it has zero idea about how to QUICKLY (in one turn) make and exploit a hole in your lines. Thus, the whole axis stratategy for 1941 and 1942 becomes impossible to make for the ai, so the game becomes very unchallenging. Giving the ai bonuses just fixes a bit of it. The ai doesnt improve at all with bonuses, it just makes it harder for you to stop the sluggish axis march and to attack axis forces later.
The basic problem for having a decent game against an axis ai is that it has zero idea about how to QUICKLY (in one turn) make and exploit a hole in your lines. Thus, the whole axis stratategy for 1941 and 1942 becomes impossible to make for the ai, so the game becomes very unchallenging. Giving the ai bonuses just fixes a bit of it. The ai doesnt improve at all with bonuses, it just makes it harder for you to stop the sluggish axis march and to attack axis forces later.
-
- Posts: 6948
- Joined: Thu Jan 17, 2002 10:00 am
- Location: Cottesmore, Rutland
RE: Slightly OT: AI Rant
My rant was not aimed at WitE specically but modern games in general, WitP is another I like to play but the scripts in that that are equally poor although some attampt has been made to improve them. Surely all the minds out there could come up with a way of creating a semi decent AI.
RE: Slightly OT: AI Rant
It comes down to money. Chris I think you and I were born 25 years too soon to have the AI opponent we want.
There are generic type AIs, like Alpha Zero, that basically learn by playing a game hundreds of thousands of times (in a short time span). I am sure it could be adapted for Wite, but why would they? I think that is the same one used for DOTA, a game you would never think AI could beat the best humans, but it has.
But for an incremental improvement in the AI, like a poster above said, a really good AI would cost as much as the core game. How many copies would sell at $200 a pop? It does suck, but it is what it is.
There are generic type AIs, like Alpha Zero, that basically learn by playing a game hundreds of thousands of times (in a short time span). I am sure it could be adapted for Wite, but why would they? I think that is the same one used for DOTA, a game you would never think AI could beat the best humans, but it has.
But for an incremental improvement in the AI, like a poster above said, a really good AI would cost as much as the core game. How many copies would sell at $200 a pop? It does suck, but it is what it is.
-
- Posts: 6948
- Joined: Thu Jan 17, 2002 10:00 am
- Location: Cottesmore, Rutland
RE: Slightly OT: AI Rant
ORIGINAL: AFV
It comes down to money. Chris I think you and I were born 25 years too soon to have the AI opponent we want.
There are generic type AIs, like Alpha Zero, that basically learn by playing a game hundreds of thousands of times (in a short time span). I am sure it could be adapted for Wite, but why would they? I think that is the same one used for DOTA, a game you would never think AI could beat the best humans, but it has.
But for an incremental improvement in the AI, like a poster above said, a really good AI would cost as much as the core game. How many copies would sell at $200 a pop? It does suck, but it is what it is.
You lost me at DOTA?
-
- Posts: 904
- Joined: Fri Oct 07, 2016 7:15 pm
- Location: England
RE: Slightly OT: AI Rant
Dota is a competetive Esports game with a big playerbase
In laymans terms " enough money for AI" [:D]
In laymans terms " enough money for AI" [:D]
RE: Slightly OT: AI Rant
Sorry- ST summed it up well. Defense of the Ancients, a 5 vs 5 game based off a Warcraft mod.
-
- Posts: 904
- Joined: Fri Oct 07, 2016 7:15 pm
- Location: England
RE: Slightly OT: AI Rant
Warcraft 3, oooooh they were the good old days! [&o]
RE: Slightly OT: AI Rant
ORIGINAL: AFV
It comes down to money. Chris I think you and I were born 25 years too soon to have the AI opponent we want.
There are generic type AIs, like Alpha Zero, that basically learn by playing a game hundreds of thousands of times (in a short time span). I am sure it could be adapted for Wite, but why would they? I think that is the same one used for DOTA, a game you would never think AI could beat the best humans, but it has.
But for an incremental improvement in the AI, like a poster above said, a really good AI would cost as much as the core game. How many copies would sell at $200 a pop? It does suck, but it is what it is.
Considering the absorbing nature of wite, I thinka good number of us would pay 200 bucks for a good ai. I would with no doubt
RE: Slightly OT: AI Rant
Considering the absorbing nature of wite, I thinka good number of us would pay 200 bucks for a good ai. I would with no doubt
AFAIK 2by3games are only 4 people working on 2 new games. They don't have capacity to work on AI like google or whatever corporation have. I really doubt 200 euros/dollars for game can cover development.
-
- Posts: 6948
- Joined: Thu Jan 17, 2002 10:00 am
- Location: Cottesmore, Rutland
RE: Slightly OT: AI Rant
ORIGINAL: AFV
Sorry- ST summed it up well. Defense of the Ancients, a 5 vs 5 game based off a Warcraft mod.
Not for me
-
- Posts: 6948
- Joined: Thu Jan 17, 2002 10:00 am
- Location: Cottesmore, Rutland
RE: Slightly OT: AI Rant
ORIGINAL: No idea
Considering the absorbing nature of wite, I thinka good number of us would pay 200 bucks for a good ai. I would with no doubt
I might now I've played it but I wouldn't before playing it. I'd look at cost and think, 'that's expensive' or words to that effect.
RE: Slightly OT: AI Rant
I was reasonably satisfied with the AI, at least the Russian. I don't expect a lot, particularly for a game made a decade ago. But I felt the Soviet AI reacted well to my moves, they took up sensible positions. In my first German campaign I was holding the winter line positions in spring of '42. During the mud season I wanted to launch an attack to capture Tambov, eliminate enemy units, and relieve the pressure being exerted on Army Group A at Rostov.
It all had to come off in a single turn because of the mud. Three panzer corps struck the line, opening a 300 mile-wide gap, and the 'figure 8' kessel trapped 25 infantry divisions. A spectacular success. But what impressed me (and I noted this at the time) was how the Russian AI reacted. They slammed in to the right flank of this penetration with strong armor and cavalry forces, and then managed to erect a new line to paper over the advance. It wasn't left as a wide open breach. The AI reacted well and properly to the events.
This attack is described in more detail here
http://simhq.com/forum/ubbthreads.php/t ... ost4451803
I found the German AI far less impressive when playing as the Soviets, as they really struggle to mount an proper offensive after the first week or two, and certainly are incapable of proper breakthrough exploitation. The AI ripped me apart in the center and north on the first turn, but did poorly in the south, allowing the extrication of virtually all of my good armored and mechanized units in Ukraine. They took Lvov first turn, but no deep move or encirclements. After this opening (which is scripted?) the German AI accomplished little, getting no further than Smolensk, which held, and then the AI was pushed back until Berlin.
So defensively, the AI can put up a good game in my opinion, but are not good at offensive operations, especially exploiting breakthroughs.
It all had to come off in a single turn because of the mud. Three panzer corps struck the line, opening a 300 mile-wide gap, and the 'figure 8' kessel trapped 25 infantry divisions. A spectacular success. But what impressed me (and I noted this at the time) was how the Russian AI reacted. They slammed in to the right flank of this penetration with strong armor and cavalry forces, and then managed to erect a new line to paper over the advance. It wasn't left as a wide open breach. The AI reacted well and properly to the events.
This attack is described in more detail here
http://simhq.com/forum/ubbthreads.php/t ... ost4451803
I found the German AI far less impressive when playing as the Soviets, as they really struggle to mount an proper offensive after the first week or two, and certainly are incapable of proper breakthrough exploitation. The AI ripped me apart in the center and north on the first turn, but did poorly in the south, allowing the extrication of virtually all of my good armored and mechanized units in Ukraine. They took Lvov first turn, but no deep move or encirclements. After this opening (which is scripted?) the German AI accomplished little, getting no further than Smolensk, which held, and then the AI was pushed back until Berlin.
So defensively, the AI can put up a good game in my opinion, but are not good at offensive operations, especially exploiting breakthroughs.
Animals flee this hell. The hardest stones cannot bear it for long. Only men endure.
-
- Posts: 30
- Joined: Sat Mar 02, 2019 10:25 am
RE: Slightly OT: AI Rant
ORIGINAL: Chris21wen
When I first got into computers I did so because finding a board game opponent was as rare as rocking horse defecation. I read an article, think it was old S&T mag, in the early 80 about computer gaming and thought that looks like what I need so bought myself an Apple ][e complete with a few 5¼ floppy discs. This is it I thought the future in gaming.
Two games from the mid 80, Gary Grigsby’s Kampfgruppe and War in Russia spring to mind immediately. Both classics and considering the size of the computer memory then (128K in the 2e) the AI played reasonably, they had too as there was no email or internet. Yes there was such a time. Not all computer game were good then some real dogs where the AI was almost no existent.
As memory got bigger games got more complicated costing considerable more to develop. Games became prettier (seems to be the main criteria for a successful game these days), certainly play well against another human player, most lack fatal errors that made some games unplayable and they are all definite more stable. What they do lack is anything resembling an even poor AI certainly nothing that matches an old game’s AI, but I’ve got a very leaky memory or should I say selective memory according to my wife.
When are designers, programmers et al going to remember all those people who do not want to play another human but the AI. Spend some money on developing a decent AI please and not those scripted ones, but something that can make a decision based on the tactical situation.
I disagree. I will never touch a game that is not visually appealing. And making the Ai tough against human is easy. Let them have more unit. If you are finding the game too easy on very hard then try the impossible difficulty. If that is still too easy, then put on some self restrictions, such as not doing anything for the first 5 turns. And try survive then. So many ways you can make the game tough for yourself. STOP COMPLAINING SPOILED KID!
-
- Posts: 6948
- Joined: Thu Jan 17, 2002 10:00 am
- Location: Cottesmore, Rutland
RE: Slightly OT: AI Rant
Posted twice
-
- Posts: 6948
- Joined: Thu Jan 17, 2002 10:00 am
- Location: Cottesmore, Rutland
RE: Slightly OT: AI Rant
ORIGINAL: Iamstrategygame
ORIGINAL: Chris21wen
When I first got into computers I did so because finding a board game opponent was as rare as rocking horse defecation. I read an article, think it was old S&T mag, in the early 80 about computer gaming and thought that looks like what I need so bought myself an Apple ][e complete with a few 5¼ floppy discs. This is it I thought the future in gaming.
Two games from the mid 80, Gary Grigsby’s Kampfgruppe and War in Russia spring to mind immediately. Both classics and considering the size of the computer memory then (128K in the 2e) the AI played reasonably, they had too as there was no email or internet. Yes there was such a time. Not all computer game were good then some real dogs where the AI was almost no existent.
As memory got bigger games got more complicated costing considerable more to develop. Games became prettier (seems to be the main criteria for a successful game these days), certainly play well against another human player, most lack fatal errors that made some games unplayable and they are all definite more stable. What they do lack is anything resembling an even poor AI certainly nothing that matches an old game’s AI, but I’ve got a very leaky memory or should I say selective memory according to my wife.
When are designers, programmers et al going to remember all those people who do not want to play another human but the AI. Spend some money on developing a decent AI please and not those scripted ones, but something that can make a decision based on the tactical situation.
I disagree. I will never touch a game that is not visually appealing. And making the Ai tough against human is easy. Let them have more unit. If you are finding the game too easy on very hard then try the impossible difficulty. If that is still too easy, then put on some self restrictions, such as not doing anything for the first 5 turns. And try survive then. So many ways you can make the game tough for yourself. STOP COMPLAINING SPOILED KID!
I beg your pardon?
Did you not read what I said, it's about the ability of the AI in computer games and if you have no valid argument as to why it has not been improved since I started playing computer game in the 80s please keep your personnal insults to yourself.
Oh and the 80s should give you some idea as to my minimum age at least, 2019-1980 = 39 but then I wouldn't be playing as a baby so lets be generous and add 15. So minimum age 54 no kid then.
RE: Slightly OT: AI Rant
Just give 'em more units! That'll fix it
Clearly he missed your point Chris.
I'm of a similar vintage, and agree that AI has failed to keep up with tech advances.
I suspect it's a combination of how difficult it is to do well and that AI doesn't sell games.
Clearly he missed your point Chris.
I'm of a similar vintage, and agree that AI has failed to keep up with tech advances.
I suspect it's a combination of how difficult it is to do well and that AI doesn't sell games.
Animals flee this hell. The hardest stones cannot bear it for long. Only men endure.
RE: Slightly OT: AI Rant
Fake AI difficulty blows, utterly. I don't want to have to face endless hordes of mindless AI troops, all doing the same exact thang, over and over--I want to see historically [when applicable] appropriate enemy forces, reacting intelligently, adaptively, and realistically to my moves. The easy way out, the cheating AI, pretty much ruins any historically-based game because it requires your tactics to deal with the AI to be warped all out of recognition from what a historical commander would have done.