Songs of Conquest - What Can The AI Do Better?
Вставка
- Опубліковано 7 вер 2024
- ADM's Article mentioned in the video:
steamcommunity...
The AI in this game has come a long way. Many players now think that it is quite strong. It's all thanks to the Captain Gars and the rest of the devs for their hard work!
However, there are some who want the AI to get even stronger! This is why I made this video.
Since harder AI difficulties only get bonus income, xp, and movepoints, I am playing with vision against a Worthy AI that does not have any bonuses just to find ways for it to improve.
The first weakness is that the AI does not have knowledge about the RMG layout and does not aim to get an expansion as fast as possible. Highly competitive players would try to expand as fast as possible because the advantages we get from the second wielder is phenomenal. We have x2 move points, more units, more scouting capacity, etc.
The second is that the AI picks up every object in its path, instead of selectively picking up stuff that are important to execute a certain plan. In my play, try to precisely calculate what I need when I plan to go for Artificer rush and only pick those things up. I plan ahead to get enough gold/basic resources for hiring wielding and upgrade a city. This way I save up a lot of move points.
The third is when the AI walks back and forth, wasting a lot of move points. It should have a clear goal of what to do, for example: It might say "I want to expand as fast as possible" and "I want to level up my town as fast as possible" because I know how much advantage that brings with additional wielders and income. With those goals in mind, it should execute their moves accordingly. Compared to my play, the AI progresses way slower. The second or third wielder should be the one picking up stay resources and the main wielder should be rushing forward.
The forth is that they are not going for the meta wielder / unit composition. In this game I manually selected the mirror match up of Everthink VS Everthink. If the devs want higher difficulty AIs to not rely too much on bonus resources, they should have them go for the meta wielder and units. In this case, Everthink Destruction rush and then Artificers and corresponding upgrades. In my play, I try to precisely calculate all the required resources to execute the meta so I would not waste turns waiting to find resources.
The fifth would be the Combat AI, by which I have not yet analyzed. If you look at my quick combat in the later part of the game, the AI seems to do very well. I barely lose units when I had a bunch of Artificers. The AI does not seem to choose the right units to combo with the right wielder. Check out this image of what the AI has toward the end of the game:
drive.google.c...
With Everthink, he went Position 1, and Destruction 3. For units, he does not focus on generating enough Destruction at all.