Poll #30: 07/09/02 - 08/20/02


Using Those Resources
Are there any AI technologies that you once considered 'too expensive' which you're now looking at using? Which ones? [171 votes total]

I've been experimenting with Learning capabilities. (31) 18%
I've been building a Pathfinder From Hell--multi-layered, multi-capable, multi-everything! (9) 5%
I've been experimenting with Neural Networks. (33) 19%
I've been building AIs that make use of genetic algorithms. (22) 13%
I've found Level of Detail (LoD) AIs to offer some significant promise of late. (8) 5%
I've been exploring some other kind of AI technology that you really don't have listed here. (25) 15%
Mostly I've been working on adding capabilities for the player to roll in their own AIs via 'bots or the like. (10) 6%
I've done lots of experimentation, but I can't really say any of it has been all that successful. (17) 10%
Can't really say as any of these choices apply... (16) 9%

Steve's Comments: I'd wondered about this ever since the editor of Game Developer suggested it as a topic I try to cover in one of my "State of the Industry" articles. Very interesting results.

At a guess it doesn't appear that any one technology is being sought after by developers as they gain more and more CPU resources--there's some fiddling with learning and Neural Networks, but nothing terribly dramatic in my opinion. The low numbers for Pathfinders are interesting--it would appear that most developers have a pathfinder that works just fine, thank you, and they don't feel like tweaking or reinventing it.

These numbers seem to match mroe or less what I was seeing at the GDC 2002 AI roundtables. Developers at this time are doing more focusing on making their tools better than they are on exploring new technologies. Neat. I wonder if this will be a trend, or if it's a temporary thing as the industry consolidates and digests its current technology?