GameBanshee Forums

Hey everyone out there.. after a long nap, I've returned to action. Anyways.. My question/topic concerns the implementation of bots/bad guys/ monsters in games that are going to come out. How much longer can we program games to get smarter and to what capacity could we to implement an AI that is rather human like. Already there exists games that learn from playing you. IE: they will try to deduce patterns about your playing style and adjust their strategy accordingly. Unfortunately many of the games that are flooding the market today there AI implementation consists of basically 'cheating' IE: giving the computer more resources / abilities / hp..
So after this long winded rant I am curious as to what you think.

Did anyone play SS2 are you suggesting we are going to have another Shodan in our midst

The BG2 AI was not exactly brilliant, most of it was down to the computer reacting instead of being pro-active, i think this is a possibility for AI in the future a creature actually being pro-active.

By the Rev

Unfortunately many of the games that are flooding the market today there AI implementation consists of basically 'cheating' IE: giving the computer more resources / abilities / hp..

I have to agree. I started to play Fate of the Dragon and within 30 minutes of playing this game I had the A.I figured out.To try to make up for this the programmers let the "computer players" have more men at the beginning.

A lot of strategy games and RPG's (these are the type of games that are suffering the most from bad AI) have a multiplayer function built in; and almost every new game in these genres will have one too. I think this is the trend/remedy for bad AI: just play against a human opponent.

Only the UT bots have been anywhere near realistic, but look how much patching that took, although i could complete the game on Adept, i got to Overlord on Masterful and low and behold the Bots are suddenly overly good and pinpoint accurate, it lost it's authenticity of skill.

Omar writes:

A lot of strategy games and RPG's (these are the type of games that are suffering the most from bad AI) have a multiplayer function built in; and almost every new game in these genres will have one too. I think this is the trend/remedy for bad AI: just play against a human opponent.

Trend, yes; remedy, I would respectfully disagree. There remain quite a few people who would rather play a game at their convenience, for as long as they want, which you really can't do in a multiplayer session (unless it's actually an online multiplayer environment, and that brings a whole new series of problems). You want to try three different courses of action with your main character? You need to pick up the kids from school, or drop everything for a sudden study date with your girlfriend? Standalone games are designed for play and replay at leisure. There's no replay in online games, and the need for a full session precludes much in the way of leisure.

There's also the technological hurdle to multiplayer internet games. Part One: despite all the talk about cutting edge modems and 64 MB video cards, many people still use 28.8 bps modem and 8 MB cards. They're not the intensive online gamers, so it's easy to overlook them--but they're real, they're there, and they're a very large bunch of gameplaying folks.

Part Two: the internet is getting increasingly constricted and busy these days, and although cable (which I use) and other high end means of transmission provide an answer to the problem, it's a shortterm one that doesn't really deal with the root of the problem. Just my POV, but I think that until the whole system is analyzed and lag addressed at the roots, the internet is going to be a problematic multiplayer resource.

Omar, I would say you're right that internet multiplayer presents a possible trend. Some games have been developed with virtually no standalone component--but they're also backed by an enormous investment in time, money and manpower online. As some companies have begun to discover, it's actually cheaper to develop a standalone game. True, you miss the longterm cashflow that comes from a successful, ongoing online multiplayer environment, but you also run far less risk of a catastrophic failure that sucks billions of dollars and creative energy into a black hole. (Sega's HEAT, anyone?)

I do think that multiplayer gaming has had a beneficial effect on standalone game AI, because basically players are now comparing AI to the level of expertise a savvy human opponent can provide. AI programmers like Sid Meiers (whose mom, by the way, used to bake the best chocolate chip cookies in the world) can command salaries and authority that simply were not available to them ten or fifteen years ago when the games were simpler, and AI was put to much smaller tasks.

Weasel's comments about Fate of the Dragon are very much to-the-point. Savvy players expect more, and look for holes in computer logic. Despite what some in the industry may believe, they don't just look for great graphics are fantastic cutaway sequences: they want topnotch AI, good game balance, a strong interface, an interactive environment and lots of gameplay elements. It's not so much the hardware, IMO, that drives the gaming software, as the players who are an older and more sophisticated group than they used to be, and have developed a series of expectations.

Whew. Sorry for getting so longwinded.

[ 04-11-2001: Message edited by: fable ]

@ fable: I think you are right about that multiplayer thing. Multiplaying can only replace computer opponents when you have the time and money to invest in a game after you have beaten it in singleplayer mode.
What I'd like to know is this:
1.did anyone of you play a RPG which presented a significant challenge because of its AI? Or lies the challenge in the fact that computer enemies are simply stronger/better equipped?
2.same question,but for BGII: was this game a real challenge or had you figured everything out the first couple of days of playing it?Is the fact that you can solo through the game a plus point for BGII or not?
3.is AI something that is limited by the technology/hardware that game developers use or do they make their games/RPG's easy so that more people(kids) can play them?

1. The AI is easily beatable i nany RPGi have played, granted my scope is not excessive the only games i can think of is the Looking Glass trio, Thief 1/2 and System Shock 2, their AI sometimes suprised many players, the only thing they missed was climbing up stairs. The better equipped issue always annoied me, OK so you are the hero and you are supposed to be the underdog, but sometimes it is ridiculous, Severance did a good job of weapon balancing, only the end boss was a bit over the top with his weapons, the Boss you meet before him is your match and this is what i like to see he had the same skill as you and more or less the same weapons.

2.The real challenge in BG2 was adapting to the story and Character progression the fighting is a side note in this and many RPG's (or at least it should be )

3.Probably this is down to Production companies as well, they don't want a game that can not be played on a slow machine (decreasing sales), Time is also an issue, look at black and white and how long that took.

Good question, Omar. I think the quality of the AI may be due in inverse proportion in part to the complexity of subsystems within a game, and the narrowness of human responses. The AI is Chessmaster 7000, for instance, is simply fantastic. It's been rated at strong International Chess Master level, when allowed to play at full throttle; and that's despite the complexity of chess, itself. But everything is on the surface: there are no subsystems to integrate within the game.

The best AI I've otherwise encountered was probably in in the Civ/SMAC games. It makes mistakes and it can act strangely, but by and large it's quite good. I admit, I have to handicap myself to create a challenge, but that's only because I'm so familiar with the game conditions and can adopt to them, while the AI is (after all) very rigid.

Also extremely good is the AI for the Creatures series, and the creature AI in Black & White. (The village AI in B&W, however, is absolutely braindead, and one of the main reasons I'm no longer playing the game.) But these are essentially toys, rather than games. Molyneux has I suspect tried to adopt Creatures' style AI into a larger gaming framework, and somewhat succeeded but (typically) overdone the micromanagement as he generally does to prevent you from strategizing too deeply.

[ 04-12-2001: Message edited by: fable ]

I agree with the Black and White comments, the creature is an advanced tamagochi, but the micro-management is way over the top, "must build more homes" tends to get extremely frustrating.

The original CIV had reasnoably AI the Following ones have developed a better perspective on the treatment of AI, i remember in CIV how i managed to control Europe entirely within the first few turns, make a treaty with the first village you see, research chariots break the treaty, keep the treaty making treaty breaking method to every village/town/city and you will eventually own europe. What Fun!