the AI are really stupid, even if they are expert

[tOXIC OVER 9K]

Please make them STRONGER!

[SMG Studio]

It will be done.
Keep in mind we have still many players who struggle to beat an easy AI...

[tOXIC OVER 9K]

OK, I hope you bring an update pretty soon.

but players don't learn with bots that are worse than yourself ;)
especially not if you call him an expert :D
the players are also smart enough to use this AI to their advantage.
I believe that the ai difficulty setting has no effect.

if the ai should work at some point, it's a good game. wish good luck

[Strossi]

At the end of the day, this is a RNG based game. One should not learn vs the ai, but straight into the fire. You will win some, you will lose some. BUt the point is, you will learn the meta, which you will never learn playing the AI.

And the meta seems very boring; stack armies and win by attrition on turn 50.

[SMG Studio]

To be fair (to our devs who try to make the AI smart) if the AI persona it picks goes against your style it will be harder. But we havent coded Deep Blue. Some players complain it's too easy no matter what. Others complain it's too hard. Some say the AI cheats (it doesnt) so it's really hard for a game with RNG and also so many variables. Compared to Chess, RISK is much harder to code AI for.

[tOXIC OVER 9K]

A good start would be if the AI, for example. attacks a continent that has been taken by someone who is far outnumbered at the borders (e.g. 10 against 1). the ai logically prefers to take over its own continent, but does not pay attention that it can easily rob someone of its continent bonus.

and some players take this risk and use it as an advantage to strengthen themselves against the really difficult borders against players and leave the border to AI largely unattended

that's to be exactly what bothers me most about AI
the player can use the AI as a defense


to the difficulties:

unfortunately i don't know exactly how the difference between the difficulties is. I couldn't make a difference between the lightest and the hardest and left it to expert.

I usually know the difficulties in such a way that there is a defensive, neutral to offensive way of playing for the AI. and the defensive stage is rather difficult, since the AI mostly conquers in slow steps (e.g. only one country or only the weakest targets in the actual plan) in order to keep losses to a minimum. offensive art occupies many countries but is weakened by the fighting. neutral is a mixture of the two

[Ethan]

It has been four years and nothing seems to have been done about this. The expert AI should be hard as players new to the game should use the beginner AI.

It is nice in many games when you can add a bot as a buffer when playing with other players or when human players are not tolerable. I would agree with the statements above that a power assessment should be done so that the AI is more likely to try and stop other players from getting bonuses and have some capacity to work with other players when someone has become too powerful. Perhaps have AI's switch modes depending on their score in the game. If they have few territories it would be good for them to be defensive, if they hold a number of continents it should be more aggressive.

[Thurston Forbush III]

Originally posted by Strossi:

And the meta seems very boring; stack armies and win by attrition on turn 50.

And they wonder why they are losing players at a rapid clip.

[Loxodon Historian]

Of all the weaknesses of the AI, the one that bothers me most is that it's rarely possible to play a good game of Capital Conquest with AI because the AI usually doesn't seem to value capitals.

I get that it's challenging to program AI that can account for a lot of different factors in deciding where to placing troops or when and where to attack, but it can recognize the value of taking and breaking continents. Programming AI to value taking and holding capitals seems like it should be relatively easy because you just tell it to place more value on territories with Capitals. Once in a while the AI will stack on their capital, but it's rare. Against 5 expert AI a few weeks ago I think exactly one bot was cap stacking.

[keedor]

I have found that the AI has problems with the Speed Blitz games, also. I have planed many of these and never managed to lose a game yet!

Usually the AI tends to be very cautious and often will not attack neighbouring territories even with a vast numerical superiority, then transfers troops away from the front lines to places where they are not needed.

Obviously the AI does not understand that this version of the game requires rapid expansion to achieve victory!