New Game AI Poll

FPS all the way, just look at the leaps and bounds they are coming suddenly. The first true AI in FPS games was Quake, then quake 2 added sound-based hunting and now bots are beginning to challenge the great multiplayer tradition in difficulty. Also, i really don''t see enough opportunity in other genres for AI improvement.

-Run_The_Shadows
-Run_The_Shadows@excite.com

run the shadows, you don''t see any room for improvement in other genres? have you played anything other than quake and it''s spawns? every game can use improved ai, bar none. i do agree with you about which genre will see the greatest improvement, fps, that is...

<(o)>

Hey Steve,

Interesting survey.

IMHO

A big benefit for AI by way will be TcL. Games that are highly graphical (FPS) can spend more than 5% of the CPU on AI. Games that usually spend more time on AI can start to be more graphical (RTS). Pretty much any real-time genre will benefit, especially in single player. More games will start to be able to use real AI instead of just faking it

An interesting place for AI and A-life would be unmassively multiplayer RPGs. Imagine playing an RPG with a few friends in your party (LAN/internet), or alone... and having NPCs that act really quite smart (so deep of characters, and seemably intelligent reactions that you''d hardly know the difference between it and MMORPGs, except less lag and less in the way of jerks).

- n8

nathany.com

Having had some peeks into numerous projects in early development I am getting the feeling that the first person shooter is about to stagnate. The latest advancements in the area have been primarily with rendering and networking (computer, not neural). Even the recent Unreal Tournament is merely extending Polge''s fundamental AI, not really an advancement in the technology.

Many companies are still clinging to "deathmatch" and "capture the flag" standards or transmuting the same to fit military analogy. But the genre born with Wolfenstein and Doom is already tiring; also, there is not great call for further advances in AI to handle the game tasks these games call for. Given a few more generations I feel confident that the market will move to a new level, and there are developers who also foresee this. More and more, 3d game developers are leaning toward role-playing aspects as the field of promise.

Some have seen the financial appeal of service games (typified by the massively multiplayer online role-playing game; MMORPG) and the number of these in development has skyrocketed. It''s good business, extending the profit potential and the sales window of a game. But these service games tend to avoid complex AI in favour of player-driven action. That may change, but probably not soon.

Smaller scale CRPGs, however, are putting a lot of consideration into AI. The appeal of licensed technology for rendering (engines developed in the fps genre like those of Quake and Unreal) is attractive to such studios, allowing in-house resources to be fucussed upon things like AI. One upstart CRPG project I came across even proposed simulated crowd behaviour (an area of AI that I haven''t seen proposed outside raw simulation).

So, it seems to me that the CRPG or Hybrid 3d Action/RPG area holds most promise for AI advances in the mainstream of games development. More experimental research will of course, probably continue within text adventure systems and pure simulations.


Something like three out of four games started never see completion according to an article I read in a recent magazine. With that sad consideration it is difficult to make any sort of prediction, much less an accurate one, from a look around at the projects in development; but maybe I''ll get lucky on this one

Hello Again:

Thanks for the kind words on the poll, nathany. I''m interested to see if this year''s results match last year''s and what I got directly from my fellow developers at the GDC AI roundtables this year.

One very interesting tidbit out of the roundtables this year (which I expand on in an upcoming Game Developer article) is that most of the developers actually thought they had *plenty* of CPU! Nobody was really complaining that they didn''t have enough, and a couple actually said that they were the ones in control of the CPU budget, not the graphics guys! Pretty amazing turnaround from a couple of years ago.




Ferretman

From the High Mountains of Colorado

Just thinking... maybe the next great AI will be on the player''s side, not the enemy''s. Wouldn''t it be great to have a realistic companion in an adventure game, or a more intelligent wingman in a flight sim?

lntakitopi@aol.com | http://geocities.com/guanajam/

I agree with SHilbert an intelligent companion would be cool and a great way to make use of more advanced AI and at the same time making the gaming experience deeper and more believable

cheers

Steveo



Edited by - Steveo on May 18, 2000 5:44:51 AM

Intelligent companions or wingmen were a big topic at the 1999 GDC, SHilbert. We didn''t talk about it all that much in any of my sessions this year though...not sure if that means developers have lost interest, everybody''s already doing it, or nobody wanted to reveal some "strategic advantage" for the next batch of Christmas games.



Ferretman

From the High Mountains of Colorado

Given the recent success of The Sims, I would hace to say that the most likely place for games to develop (in all areas, not just AI) is probably the Simulation/Strategy front. Picture a town such as New York, filled with millions of people. Now say that GodZilla attacks the town. In the average game today, what you would get would be a bunch of people standing around going "oh gee, that big monster is coming" and pacing in a 5x5 room. In a short while, I think we will see more realistic response, such as people running, screemin, looting, and military intervention similiar to the actual GodZilla Movies. The people would actually react to things that you do, such as jumping in the city fountain. This same type of AI will probably be implemented in RPGs as well. About the Smarter companions- I think I would like to see smarter REAL players more than seeing smarter computer controlled players.

Etnu

What is a man without goals? A dead man.

I tend to agree with you a bit,
Etnu, though I think advances will most likely come out of the big online multiplayer game arena. (The simple fact is, that''s where a heck of a lot of money is being dumped over the next couple of years.)

I''m finding the results of this year''s poll on the subject interesting. Here are the numbers as of 9:15PM Thursday night (5/17/00) compared to the original poll taken during the month of May, 1999:

Where do you think the next innovations in game AI 
will come from?
            
                                2000 Poll       1999 Poll

First-person shooters?             19%              24%
Real-time strategy games?          20%              38%
Turn-based strategy games?          5%               9%
RPGs?                              18%              22%
Sports games?                       1%               7%
"Classic" games of some kind
 (checkers, chess, etc.)?           2%              N/A
The academic world?                 4%              N/A
The military research world?        9%              N/A
I don''t forsee any real innovations 
coming for quite some time yet.     6%              N/A
Other (type in your own genre).    10%              N/A

Note that while I didn''t offer all the choices last year that I did this year, the percentages of the groups that I did offer are roughly comparable, particularly if we allocate the "other" and "none" categories even across the original five choices. Sports and RTS games seem to be the "winners" so far, however, with fairly large increases from last year.

Personally that surprises me a bit, I guess....RTS games don''t seem to require the same level of interaction for success as other genres (note that I didn''t say they didn''t need it, just that it didn''t seem necessary for market success). Sports games I can kinda understand, I guess, since the developer is driven by highly knowledgeable players of the "real" game in that respect.

Interesting numbers, nonetheless.





Ferretman

From the High Mountains of Colorado