How far is the AI right now?

I just got 8 hours into the alpha and found it pretty ok for an early alpha actually. Most stuff works, what does not seems to be disabled. Only one thing really bothered me. The absurdly dumb AI. 2 out of 3 opponents never cared to attack me and where so far behind in all aspects of the game that i did not even have to use any fighting vessels to beat them. Just send in (slow!) troop transports and took their worlds. The last opponent had about 150 small vessels that all died to one large ship of mine.

Looking at the planets i took it was even worse. Food production on research bonus fields and said research buildings where sitting on food bonus tiles.

This leaves me to the conclusion that the AI currently is not even activated at all apart from some very basic "do something random".

For features we have this topic: https://forums.galciv3.com/452714/page/1/

Can we get a similar topic for the AI? Or did i miss that? AI is one of the, if not the most important part about these kind of games. If you implement a new algorithm, LET US KNOW so we can test it as directly as we can. But we need to know if something changed and what exactly. There need to be algorithm for efficient planet building, shipdesigns, attacks, defences,... reactions to attacks, reactions to attacks with specialized vessels ect.

 

Thanks.

4,160 views 7 replies
Reply #1 Top

The current AI is a place holder. It was "updated" to attack but is still completely stupid as to allow the players to test other features of the game.

Reply #2 Top

I see. But... why?

AI is the hardest thing to get right. It should get the most testing. It is easy to write AI that beats every player, everytime. But it is hard to write it just good enough to be a challenge, without doing too many mistakes. That takes testing and even more testing.

 

All the other features can, and should be tested meanwhile. Not exclusivly. :-(

Reply #3 Top

Quoting Zaradoom, reply 2
I see. But... why?
End of Zaradoom's quote

Because a lot of the game features are either not implemented and/or finished yet. If the devs start to write the AI now, the'd have to re-write it every time a new feature gets added, or a existing one changes. That would increase the workload immensely.

Reply #4 Top

As a Codemonkey myself i can savly say: "no additional workload". Well, technically youd have to comment some lines out,... that does not count as additional work tho and is assuming you get the AI right on the very first try, AI is something you have to adjust SEVERAL dozen times anyways. Might as well do it stretched out.

Reply #5 Top

This is a big reason why I'm waiting to play this.  

 

I'm confident Brad will do a good job, but I know the game won't be fun until it does- and unfun isn't really my thing.  I put the $100 down because I think it will be a good deal in the end, willing to wait for it.

 

Reply #6 Top

I wont be writing any of e AI until this Fall. But Alan, one of our best developers is implementing the APIs I asked him to and is putting in test values currently so it will be starting to play a decent game over the Summer.

Reply #7 Top

Hey Brad, did you get a free founders badge, or did you put the hundred bucks down...?  ^_^