Anybody who has played Endless Space is aware that the ES1 game AI isn't the most impressive. While I don't think commercial-level 4X game AI will ever be as good as players feel it should be, I do think there's endless (wink) room for improvement with its iteration in ES2. Especially considering some of the bad synergy between poor AI and unbalanced mechanics evident in the current Disharmony version (eg. long range kinetics dominance and AI ship design oblivious to this fact).



With that in mind, I have four questions. Two for the devs and two for the players.



Devs: What can you tell us about the AI design for Endless Space 2? What areas have been focused on to address AI shortcomings as found in ES1?



Players: What are the most frustrating and/or deficient behaviors/actions you've experienced with AI interaction in Endless Space? What do you think are the most essential requirements for ES2's AI to really shine?

I think the worst part is how the AI does a lot of crap that players don't do that sets them back long term and is annoying, like their shitty scout killer fleets at their homeworld, speaking of which, the AI should be less aggressive towards scoutships. Also, there can be a big jump of AI challenge between difficults; on hard, I always have like 4x the score of the AI, on serious, the AI is usually so OP(In addition to the then frustratingly high disapproval bonuses due to difficulty) that I end up ending games early. Also, the AI declaring on you randomly for a higher score is bs, especially since that usually means they lose cause you're better than them. The AIs way of having absolutely crap ships in combination with shipspam is just terrible. And the trading system is often odd; in first contact/early game situations, the AI should usually accept white peace(Rather than not accepting peace at all or giving you stuff for peace), and later on, the AI often doesn't realize it's situation and acts at random. Also, again, unreasonable war. I can't stress this enough; in a game I did today a Hissho AI who had 2 crappy borderless systems within one of my planets huge influence radiusin addition to some crappy system I occupied and gifted them for trade partners declared war on me after it had finished building 1 scout. Interestingly, in the same game a Sherdyn AI that was right next to me and doing really badly did seemingly nothing at all; at ~150 turns on normal I noticed that they were completly enveloped by my borders and lacking late early game techs like barren and lava colonization; it seemed that my borders had somehow paralyzed the AI.

I think a big step towards better AI would be artifical; making the AI do things that work well and avoid things that are crap, so that they start the game building isotope reactor refineries or food or industry exploits rather than combat ships and that they build a decent number of colony ships.

Also, it would be nice if there was a way to communicate territorial claims(Your home constellation especially and probably automatically and with extra weight) to the AI; then it would have to decide between an act of war(In the AI code, not in the actual game; so that the AI treats it as declaring war) or not colonizing that system right on my wormhole. Then again, that idea sounds kinda like it could go really wrong(Especially if the AI itself could do it).