The Game Developer’s Conference 2008 is now over, and while those who attended spend the rest of the week recovering and catching up, the rest of us can start piecing together the important bits of information that came out of the event.
Along those lines, I just noticed a report from the session about Assassin’s Creed, from the blog of veteran game developer Rob Fermier (a.k.a. Xemu). One line in particular jumped out at me:
“50 software engineers at the peak. AI was 15-20.”
Just to make sure you got that right, there were between fifteen and twenty AI programmers at Ubisoft working on Assassin’s Creed. Now, I’ve been lucky enough to work and contract in some pretty big companies — but that’s just incredible! I’m simultaneously in awe at the possibility of what so many talented people could do, and also wondering about all the different kinds of management and engineering nightmares this could cause.
Basically, that’s the topic for this Tuesday’s developer discussion here at AiGameDev.com — officially the best place to take GDC’s AI discussions further!
How Many AI Programmers Does It Take…
In the hope of making half of you feel more adequate about the size of your team compared to the average, here’s a quick poll to see where industry is at these days.











Comments
Comment on this article. | Show full forum thread.If you have a variety of possible interactions like they do in Assasin's Creed, such as crowd behaviors, guard behaviors, etc., most with associated animation controls, the numbers of tasks are going to add up quickly.
But then they employ a hell of a lot of useless programmers (i.e. a lot of graduates that cant do much useful code, but are hellish cheap).
Cant really remember the exact figure, but it was something insane.
I find this trend annoying because it makes it hard for AI programmers to focus on (what I consider "real") AI tasks, and this slows down progress on AI technology and design. I guess in the end all is well because AI is not the only area where advances are useful and worthy, but still.
For the record, I consider AI to involve complex and/or mid- and long- term decision making: pathing, planning, coordination, perception, memory. Executing those decisions is a different thing.
If I recall right, the figure was something like 150 programmers. But I could be completely wrong there. If it was "gameplay + ai" then maybe that works. I know they had 50+ for the UI and 15+ for online.
Stupid numbers and totally unnacassary, my buddy was working on a previous version of the game (about 4-5 years before) and was only on the AI for half his time. Think the players noticed a 300% improvement? I doubt it :)