Jump to content
Sign in to follow this  

information about the Ascension AI...?

Recommended Posts

greetings and salutations --


i had originally posted this over on the Strategy board, but maybe this would have been the more appropriate place for it...?  i don't know for sure, there are arguments to be made for either, or for both, so:


greetings and salutations --


so, i find myself rather fascinated by the Ascension AI, the way it plays, how it assesses such a complex game and an ever-changing board state...  my interest is purely academic, i'd just like to know more about how a machine can be taught to think and understand and prevail at such a complex endeavor...  it doesn't seem like it could simply be tallying values and making decisions moment-by-moment -- it seems too intuitive for that...


so, without giving up trade secrets, can you guys point me toward educational materials on how this specific AI, or, failing that, how AIs *like* Ascensions do what they do, what they're capable of, what their limitations are, what directions they may be moving in, and such...?


i know very little about the artificial intelligence field in general, but, as a lifelong sci-fi fan, it has always intrigued me, ever since HAL...


thank you in advance...


peace --


-- khs


p.s. - if you answer "I'm sorry, Dave, I can't do that..."  i'll just be hiding under my bed...  forever...

Share this post

Link to post
Share on other sites

Hmm well, I quite doubt the Developers will ever handle the AIs' codes to third parts : AIs are usually the most complicated feature to implement in any games. That's why many videogames were made for 2 or more players, years ago.

Implementing an artificial player that simulates real ones' actions is nothing to laugh on, indeed. I personally tested it, since I'm a newbie software developer, too. :P


I played against easy, then hard AIs, during my Ascension's playlife, and I still can tell something : it pretty seems like the easy AIs just buy random cards according to their gained Runes and Power, but they often prefer acquiring Mystics and, especially, Heavy Infantries (LoL). Something like :


Play All;

• Resolve cards actions;


Check [Rune] and [Power];

Check [Center Row];

• IF a card can be acquired from the center row and costs 6+ Runes THEN

•• Buy a random card filtered with the IF instruction;

• ELSE IF a card can be acquired from the center row and costs 2- Runes THEN

•• Buy a random card filtered with the IF instruction;

• ELSE IF a monster can be defeated from the center row THEN

•• Defeat a random monster filtered with the IF instruction;


•• LOOP :

•• Check [Rune];

•• IF [Rune] = 2 THEN

••• Acquire Heavy Infantry;

•• ELSE IF [Rune] >= 3 THEN

••• Acquire Mystic;

•• REPEAT UNTIL no more of them can be acquired;

REPEAT UNTIL no cards can be acquired or defeated anymore;


I, of course, voluntarily omitted instruction codes and variables declaration, letting anyone understand this, but I think it pretty works like this, for easy AIs, since I never noticed a delay in their reactions and play times, meaning the code had to be pretty short and easy, without any complex rooted IF, LOOP/UNTIL, CASE IF etc.


For what instead concerns hard AIs, it's pretty difficult to tell...

As everyone playing at MAX difficulty could have noticed, reaction and, especially, play times of hard AIs vary each time they acquire, defeat, draw or activate a card. This means hard AIs' codes must be abnormally huge and complex, and I believe they do examinate each card in the Center Row, the Void, in AIs' decks and AIs and opponents' discard piles before acting, just like a real player would (rather, I often found hard AIs being more intelligent than some players I know, LoL...).


I believe they act like this, more or less. I'm no developer, so I can't tell for sure. My interest's purely interest-aimed, I always loved sotwares and codes. :D

Share this post

Link to post
Share on other sites

hiya --


thanks for the insight, Noloic...  


i guess i've always just wondered if, were you to set an AI at a sufficiently complex problem, might it not, because of the complexity of that problem, figure out stuff that wasn't intended, and thereby "evolve" in some fashion...?


i guess it's an extension of the whole "you can learn the right answers until you learn the right questions" sort of thing, only applied to an other-than-human seeker of information...


or something...




peace --


-- khs

Share this post

Link to post
Share on other sites

Uhm, this is getting quite abstract... but I think I've understood what you meant.


Indeed, there are quite many situations that surely couldn't be studied in advance by the developement team, nor the AI makers, but I think they categorized some (or all) of them into many, different "reaction cases".


What an AI misses and surely can't gain, though, is the long-term thought project : an AI doesn't think like a human, and can't know what its deck is aimed to, in advance. Speaking of which, I guess the definition of "AI evolution" is quite wrong, in this sense, since they have a static thought, and they follow this kind of gameplay, since the beginning to the end. Studying an even complex AI would imply years of work and too many variables, and, even in that case, it wouldn't be sure to be an evolving AI, either.


But, I repeat, this is getting ephimereal alot, I think many readers probably don't know what I'm talking about. :P

Share this post

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Sign in to follow this  

  • Create New...