I’ve been playing a lot of Sid Meier’s Civilization recently and have thought about this conundrum: should AIs in video games play to win, or just play to make the player lose?  These are two different strategies mind you, if each AI is playing to win they will act in their own rational self-interest to pursue their own goals.  But if they are just playing for the player to lose, they may instead act against their own self-interest in order to hurt the player.  I feel like I see this “play for the player to lose” strategy a lot in games against the AI but I don’t know if it’s accurate or just my imagination.

Consider for instance the early phase of the game: settling.  The player and AIs all are scouting and settling in an attempt to claim as much land as they can to grow and become stronger.  I often appear to see AIs make baffling settling decisions, settling on terrible land with zero fresh water, and their decisions only seem to make sense if they are simply trying to box the player in, not actually win themselves.  Settling is an expensive process requiring a lot of food and production, so you want to settle as good a city as you can.  On the other hand giving the player a lot of land for themselves lets the player grow stronger and be more likely to win as a result. 

The compromise seems to be that in games I have played, the AIs nearest the player will settle on marginal lands in the direction of the player, boxing the player in and preventing them from expanding.  Other AIs will then have more room to settle good land and actually attempt to grow stronger and win the game.  In this way the game becomes more difficult for the player, even though some AIs are making choices that aren’t actually in their own rational self-interest.

To be blunt, I don’t like this.  I think AIs make the most sense when they act in their own self-interest, rather than having a secret alliance against the player in particular, I think it makes the most sense and more accurately represents how a player would play as well.  But like I said I don’t have any hard evidence that the AIs act this way, maybe they settle bad spots because they’re just poorly coded in general.  But it sure does feel like they’re all in on it.

I can tell you a game that DEFINITELY has an anti-player bias and that’s the Total War series, which is part of why I stopped playing them.  In the Total War series, every AI that borders the player in any way is just a short step away from war.  This was fine and fun in Rome and Medieval Total War, where the economics of the game made world conquests like this fun, but in Empire and later Warhammer Total war it just gets tiring and unfun. 

To give an example: Empire Total War takes place starting in 1700s Europe.  France and England both have colonies in North America, and there’s a Native American tribe, the Huron-Wyandot in central Canada.  If the player plays as Britain, this tribe will inevitably attack Britain and stay peaceful towards France.  If the player plays as France, this tribe will attack France and stay peaceful with Britain.  This again isn’t so fun.  Like I said, the economics of this game make world conquest a boring slog rather than a fun romp like in previous interactions, but also this is a historical strategy game that in certain ways does attempt to model diplomacy and agency of historically relevant peoples and nations.  Shouldn’t it be possible for France to attempt an alliance with Native American peoples to counter Britain, just as France did in real life?  I think the game would be a lot more fun that way instead of being railroaded into an “everyone against the player” scenario no matter what country you play as.

Anyway those are my thoughts on AIs, anyone else know of a game that seems to have a strong anti-player AI?