| Session Name: | Building an Open-Ended Conversation System for Expressive Game Characters |
| Speaker(s): | Aaron Reed, Alan Hinchcliffe |
| Company Name(s): | Spirit AI, Spirit AI |
| Track / Format: | AI Summit |
Did you know free users get access to 30% of content from the last 2 years?
Get your team full access to the most up to date GDC content
| Overview: | Interaction with game characters has long been striving to move past simply killing them. Static characters ruin immersion and conversation menus are clunky and repetitive, particularly in ultra-realistic contexts like VR and AR. What's needed is a more natural way of interacting with characters that is not only deep, but natural-feeling. Drawing on expertise from both AI and interactive story, this lecture describes the challenges faced when building a system for open-ended conversation. It explains techniques including classifying freeform player input, assembling naturalistic responses from underlying knowledge models, and helping writers author content for a non-linear and combinatorial system. |
Recommended
