Install Steam
login
|
language
简体中文 (Simplified Chinese)
繁體中文 (Traditional Chinese)
日本語 (Japanese)
한국어 (Korean)
ไทย (Thai)
Български (Bulgarian)
Čeština (Czech)
Dansk (Danish)
Deutsch (German)
Español - España (Spanish - Spain)
Español - Latinoamérica (Spanish - Latin America)
Ελληνικά (Greek)
Français (French)
Italiano (Italian)
Bahasa Indonesia (Indonesian)
Magyar (Hungarian)
Nederlands (Dutch)
Norsk (Norwegian)
Polski (Polish)
Português (Portuguese - Portugal)
Português - Brasil (Portuguese - Brazil)
Română (Romanian)
Русский (Russian)
Suomi (Finnish)
Svenska (Swedish)
Türkçe (Turkish)
Tiếng Việt (Vietnamese)
Українська (Ukrainian)
Report a translation problem
Discussions Rules and Guidelines
Report this post
So yeah, it's a thing.
They also have less public order maluses than the player, allowing them to tax settlements for more food than is usual for the player, who will often be on Normal Tax rates. Again, these bonuses to AI increase with campaign difficulty.
AI units may also cost less food in upkeep, again, largely dependent on difficulty setting.
Keep in mind that rebel stacks will tend to be larger, because their size scales with how discontented the original public order was, as well as the proportions of Allegiance. Thus, even a single rebel settlement may be a 20 stack, although it can also be as small as 5 stack.
AI will still have to follow rules for food consumption, just like the player. I liberated and made vassal Cumbria last campaign, for example, which is an emergent faction. They initially got 12 units when they spawned, but owing to being very poor (one region with a walled town), they gradually disbanded down to only 5 units, and have stayed with that ever since, around 25 turns later.
So it can vary, depending on regions, faction traits, leader traits, and political situation.
In general, if you lower the campaign difficulty setting then you will tend to see much less AI stacks.
I was playing Dwarves in TWW2. Had the orks to the south basically wiped out, then i notice they keep amassing these armies and hitting me from all sides. I'm like they have to have another city. After i beat their 3 full stack armies and a 'waagh' army and finally eliminate them i notice that they only had ONE territory and ONE city. This by the rules i play with as a human player is impossible. You cant support one full stack army let alone THREE with one territory. it is impossible. Not for the AI apparently.
Its the same for this game and for TW 3 kingdoms.
I don't understand how TW games are labeled as the 'pinnacle' of strategy games when basically your just playing a cheating opponent who uses no strategy and just rushes mass units at you continually even if they cant realistically support them.
I've had better AI opponents when I used to play Age Of Empires back in like 2002. no joke.
I do not understand how in 2020 STRATEGY game companies cant make a decent AI that doesn't cheat or just use 'swarm' tactics.. Like i said, older strategy games have better AI.. I mean didn't we as humans make computer AI in the 1970's that could literally outplay humans in chess and not cheat. Here we are in 2020 and we cant make a good AI for games by multi million and billion dollar companies.
COMMON.
The simple answer is that it's not as easy as you think and AI programmers make a hell of a lot more money than game coders.
- Chess is a relatively simple game where set strategies and complete "memory" give the AI an advantage over humans. And yet it was way past the seventies before an AI, using super computers, could beat professional chess players
- An individual game is just one product in a line of products with limited revenue potential- budgets are not unlimited
- Games are much more complicated than they used to be.
- Games need to appeal to players with wide variation in computer specs, the AI you want could very well make your computer obsolete.
These are good posts.
Citing computerized chess games really does tell the whole story, in more ways than one.
Not only does it illustrate the difficulty in coding AI, but modern chess games also show how the ability to make a challenging AI correlates a lot with how simple the game mechanics are.
The more mechanics a strategy game has, and the more nuanced those mechanics, the more advantage a human will have over any AI, something people should keep in mind when they demand more and more mechanics in the campaign game, or wine about the cutting of mechanics, like ambush stance, forced march, agents, region trading, etc.
All of these types of things just give the player more options, making it harder for an AI to ever actually challenge him.
Even the 3D map heavily disadvantages the AI, and if anyone doubts that, all they need to do is get MTW and load a campaign up, and see how much "better" the AI was from a seventeen (17) year old game. The AI wasn't actually more advanced, of course: the game design simply made it more practical for the AI to operate strategically, whereas today's TW strat maps are more about graphical and aural immersion, role play, etc.
So the whole idea of the AI factions being other "players" is silly, at the end of the day. The player is not playing against other players in SP...he's playing against the game, and the AI scripts are just part of that game. AI is not actually intelligent. There is no self awareness, no abstract thinking, no thinking at all, tbh. The design needs AI buffs to scale difficulty.
People tend to treat computerized strategy games like it's poker with like six guys at a table, with the player being one of the guys. What it actually is more like, is playing pinball: the player is the player; the AI factions, the dilemma events, the missions, etc. is the pinball machine; the mechanics are like the flippers, bonus score and extra ball opportunities, etc.
And it shouldn't be necessary to mod the game to lower buffs and such: that's what the difficulty settings are for. It makes no sense to mod the game to make it "fair" so someone can play on "Legendary." The whole point of the higher difficulties is to buff the AI!
To say that these million/billion dollar companies cant make a decent AI that uses more than 'swarm' or 'rush' tactics or just obviously cheats is sort of silly imo. Filling the map with tons of units and just zerging you constantly while ignoring the rules isn't strategy. I call it played out and silly. Like I said before. AI from nearly 20 years ago did this and did it better honestly.
The ability and technology is most certainly there and I find companies lazy for just putting the same old scrpited AI over and over. Even on the battle field not even on the campaign map these game have literally zero strategy for a so called serious strategy game.. the AI is so predictable even in the battles is a joke. I mean I question if there is any AI at all. It's basically just swarming you or rushing you with no 'thought' at all. A wall of units charging at you over and over.
We can do better in 2020. As gamers wanting a serious strategy game to enjoy and the money we've put into these games expecting a serious strategy component.. we deserve better too. Imo.
Compare it to a situation where you play a large number of battles against the same person. You will learn how that person approach things, what types of mistakes they tend to make, their strenghts and weaknesses. And you apply that so that you are more likely to win. But at the same time that person also learns how you do things, what mistakes you make and exploit that so both of you will be forced to adapt. While against an ai only you learn, and so as you figure out how to beat it it will never adapt, and there is no feasible way to change that.
As far as ai for older games go, no it wasn't better back then (apart from games being simpler would generally mean the ai gets by better due to a less complex environment as noted by others earlier in this thread), you where simply a less experienced player so it felt better, if you go back to the older games now and look at them honestly you will find that the ai in many of them wasn't really as good as you remember (the games will likely be fun still).
But getting to your point, there's a presupposition that this is a numbers problem, that better cpus should mean being able to code a better AI. That presupposition doesn't hold up to the reality though, unfortunately.
As Monteizo points out, the AI is not a learning machine. All AI fundamentally is, is a series of scripts and logic gates. It really doesn't take much computing power at all to understand these scripts and gates. An ancient 486 cpu could, with modding, execute the TW AI scripts, it just would take a month of Sundays to do it.
In other words, there hasn't actually been a fundamental change in what cpus do, only how they do it. In January 2020, the code is still in binary, like it was in 1945; it all still boils down to basically writing to a program, "if X= 1, then P, else Q." There is no amount of extra ghz, core count, threading, system memory, or on chip cache that can change that.
Up to a certain point in time, continually better hardware did mean the chance for better AI in games, because more code could be run in a reasonable amount of time. But we've long crossed that point, and there is a plateau.
A revolution in RTS and TBS AI now would require creating actual logic for the AI, not just running more logic gates in parallel, or running them faster. The reason that hasn't been done, is because you need actual consciousness to see actual logic, and no one has been able to create or synthesize consciousness.
Tbh, no one has been able to even empirically demonstrate what consciousness is, and if a person Googles "consciousness studies," this is a very fascinating thing. Back in the 1950s, it was taken for granted by almost every empiricist that advances in physics, neuroscience, biotech and genetic science would reveal what and how consciousness is, and soon. Here we are almost seventy years later though, and, ironically, we actually have more questions, more gaps in our knowledge, about the biology, genetics and cognitive process than we did back then, and are less close to isolating what consciousness is than ever.
Just as with physics, we've basically run into a wall, and no fundamental advances have been seen for decades.
On that note, I wouldn't hold out for quantum computing to make AI either. About 95% of the stuff you hear about quantum computing and the coming revolution in "true AI" is ♥♥♥♥♥♥♥♥♥♥♥. Media that wants viewership, and scientists who want grants, combine to stoke all of this mystique about androids, "AI" and quantum computers...the fact is, it's still ones and zeroes, it's still logic gates. A computer with a woman's face that has animatronic lips and can bat her eyes, and says "I want to learn," is still just a glorified Eniac: a team of people who *already are conscious beings* put logic gates in so that the ones and zeroes say "I want to learn" if someone asks "her" if she wants to "know" something or not, so it looks like "intelligence" has somehow materialized on a TEDx stage.
What I do expect in 2020 from a multi million/billion dollar company that is calling there product strategy or from consumers is seen as a deep strategy experience if not the Pinnacle of strategy gaming , is that. A deep strategy experience. All I've gotten from TW games is an AI that blatantly cheats and literally does nothing more than spam you with unit rush.
On the battlefield from what I've played of TW it is literally just a line of enemy troops marching at you with no strategy in mind. It is literally devoid of any strategy whatsoever..most of the time they don't even realize they are visibly being flanked and such.
I've never had a.. OMG wth did the AI just do moment in TW. Like I've never been amazed. I have been in other titles and some that are much older.
There are alot of flash to these games and you personally get to make a lot of cool decesions on a pretty looking and immersive map with cool and colorful units... When you realize though that your opponents are literally and strategically non existent my experience at least is very...lessened. I mean even some of the base trade deals and alliances are an absolute joke at best and are just mind dumbingly stupid on the diplomatic fronts..
The AI tech is out there now to make these games much more immersive from an AI standpoint. I personally believe they don't want to put forth the effort when what they have just 'works'.. As a consumer when I can play app games that out play me more than my TW games or games that are 20 years old do... Yes that's an issue in 2020.
And roughly translates to saying nothing more than, "we can put a man on the moon, why can't we <insert thing you think is so blindly easy it should have been done ages ago>".
I've heard many claims over the years about such-and-such a mod supposedly improving the battle or strategic AI, but I've not seen a bit of evidence to suggest that it actually does. I don't know how it could.
I applaud modders (I am one myself - shoutout to the Fourth Age Total War for RTW:BI), but the idea that mods can make TW AI better betrays, I think, a lack of understanding about what modders can do and how these games are actually designed.
As for AOE2 having "better AI" than TW games - again, that is to be expected. Those older titles were *simpler games*, with fewer moving parts.