Announcement

Collapse
No announcement yet.

[*Finished*] Researching AI in games

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    [*Finished*] Researching AI in games

    Hello community!

    I'm doing a research about Artificial Intelligence in games and i was wondering if you all could answer a few questions for me.

    If you do not want to answer all the questions please answer only this one: Would you like to see more documentation about Artificial Intelligence? A few tutorials perhaps?



    1 - How often do you play games?


    2 - Do you ever think the Artificial Intelligence in a game is really dumb? What game was this in and what happened?


    3 - What basic functionalities do you think are necessary in a game to make it challenging?


    4 - Does Artificial Intelligence in games in general need to improve?


    5 - Have you ever programmed Artificial Intelligence for a game? (if no: skip) Did you stumble upon any problems? What were those problems?


    6 - Would you like to see more documentation about Artificial Intelligence? A few tutorials perhaps? <- Important





    This concludes this survey. I would like to thank you all in advance for helping me with my research

    #2
    1. not that often, when im amazingly bored and have a new game.
    2. reference to the udn doc where someone says wow these ai's are dumb, and they were real players.
    3. entirely depends on the game
    4. yes, there is a ton of awesome academic research into ai, pc's are quickly becoming powerfull enough to cope with it in a gaming situation, cant wait.
    5. yes, the problems i encountered were solely due to (still O_o unfixed) bugs in udk's navmesh system, so i avoided navmesh altogether.
    6. yes docs would be great, but since theoretically ai's and the games they play are as individual as people, its a rather large feild to cover in a few documents. some kind of modular ai system would be really interesting, kinda thaught this mysterious AITree would be here by now and thats what it would be, oh well. i do my fair share of ai tutorials/examples, would be good to see some other approaches though.

    Comment


      #3
      1, most evenings if i have time
      2, yes but i always play on easy
      3, AI levels eg:- easy medium hard , don't go for God-like
      4, yes. dumb bots as co-op players
      5, no
      6, i sure it would help those people who can code

      Comment


        #4
        Thanks alot for helping me guys i hope more will follow!

        Comment


          #5
          1 - How often do you play games?
          I'd say atleast an hour a day when I can.

          2 - Do you ever think the Artificial Intelligence in a game is really dumb? What game was this in and what happened?
          Honestly AI seems dumb to me all the time. AI tends to loop so every enemy that may have been smart the first encounter, becomes extremely predictable throughout the rest of the game.

          3 - What basic functionalities do you think are necessary in a game to make it challenging?
          Depends on the game.

          4 - Does Artificial Intelligence in games in general need to improve?
          Yes, but like I said above AI seems to loop making them bad. If AIs had multiple AIs to act from I think that would be an improvement.

          5 - Have you ever programmed Artificial Intelligence for a game? (if no: skip) Did you stumble upon any problems? What were those problems?
          No

          6 - Would you like to see more documentation about Artificial Intelligence? A few tutorials perhaps? <- Important
          YES

          Comment


            #6
            Thanks alot YakZSmelk

            Comment


              #7
              1 - How often do you play games?
              2-8 Hours a day.


              2 - Do you ever think the Artificial Intelligence in a game is really dumb? What game was this in and what happened?
              Every time you understand the programming and used it against them. Like when you shoot at them they will come and after x time they will go away. When you know x you can go out of cover exactly when AI is turning. (Gothic 3 for example :P)


              3 - What basic functionalities do you think are necessary in a game to make it challenging?
              The AI do not have to do special things. A AI is good enough when I cant get behind the logic of its actions and so cant use it against the bot. If i can, all the immersion of the game is destroyed and the best ingame charr is "only a bot".


              4 - Does Artificial Intelligence in games in general need to improve?
              In FPS its very good. But I hate cheating AI like hell. In strategic games AI has to improve a lot. Today it is still cheating on ressources and doing good in fighting with no strategy but multitasking power :P

              5 - Have you ever programmed Artificial Intelligence for a game? (if no: skip) Did you stumble upon any problems? What were those problems?
              No, only worked with ActR to build approximate cognitive simulations of human behaviour.
              Maybe one of these simulations could be teached to play tetris ...

              6 - Would you like to see more documentation about Artificial Intelligence? A few tutorials perhaps?
              Yes! It would be nice to have a tutorial how to organize a "big" AI with lots of possible actions.
              Doing a action like moving, aiming or simple reacting is possible to figure out on your own. But getting all that in a systematic code...^^

              Comment


                #8
                1. Almost daily, unless it's been a busy day

                2. Easily exploitable yes, but I wouldn't define that as dumb. AI can give the illusion that they are smart(challenging) when they
                are in the correct 'field'. e.g AIs in UT for example work well on most Epic-made map, but can be easily exploitable in maps with tight corridor
                and less than ideal features for AI to use.

                3. I'm under the assumption that you're defining the term challenging as 'fun and provides a fair challenge'.
                Depends on genre, but generally what I think as ideal AI would be:
                - Sensing : AI must first sense(see/hear) what the players have before able to react to it ( no/limited cheating on intel )
                - Thinking : Should re-act based on player actions rather than a linear path. A combination of both is ideal.
                - Better co-op experience : Would really love to see more & Better AI that's able to work closely with human players on coop games.


                4. Yes for the same reasons that the above isn't satisfactory in a lot of games. Especially the co-op part.

                5. No . But I work closely with my lead programmer who's developing AI. UDK-specific wise, we find the lack of documentation of navmesh constraints to be quite troublesome.
                And in general, lack of lower level access to pathfinding capabilities in UDK. Generally very hard to add 'bias' to pathfinding calculations.

                6. Definitely.

                Comment


                  #9
                  Thank you guys, This concludes my survey. I will now collect all the answers.

                  You all helped me alot!

                  Comment


                    #10
                    Yo, that's a very short survey and only 1 person had actually coded AI before.

                    I can fill it in too if it's not closed.

                    Comment


                      #11
                      he actually posted this in several places, so these are not the only reactions

                      Comment


                        #12
                        Ok, sorry, my bad.

                        Comment

                        Working...
                        X