July 20, 2005
New Dissertations on AI-Based Interactive Art, Character and Narrative
I thought I’d link to a few dissertations that have been published recently that may be of interest to GTxA readers. Perfect for a summer read on the beach with your laptop, right? Here are excerpts from their abstracts.
Autonomous Expressionism and Network Arts: New Paradigms in Art, Emotional Interaction, and Information Retrieval, by David Ayman Shamma, Northwestern University
In this dissertation, I depart from traditional computer science metrics and methodologies and introduce a new framework for building computer systems with an emphasis on the creation of new artistic installations and interactions. Specifically, I introduce Autonomous Expressionism, an extension of Abstract Expressionism, whose goal optimizes the emotional experience in human computer interaction.
Narrative Planning: Balancing Plot and Character (pdf), by Mark Riedl, North Carolina State University
In this dissertation, I explore the use of search-based planning as a technique for generating stories that demonstrate both strong plot coherence and strong character believability. First, I describe an extension to search-based planning that reasons about character intentions by identifying possible character goals that explain their actions in a plan and creates plan structure that explains why those characters commit to their goals. Second, I describe how a character personality model can be incorporated into planning in a way that guides the planner to choose consistent character behavior without strictly preventing characters from acting “out of character” when necessary. Finally, I present an open-world planning algorithm that extends the capabilities of conventional planning algorithms in order to support a process of story creation modeled after the process of dramatic authoring used by human authors.
Hierarchical Parallel Markov Models for Interactive Social Agents (pdf), by Robert Zubek, Northwestern University
In this report I present hierarchical parallel Markov models for the creation of interactive social agents for video games and entertainment. The approach extends existing, popular character technologies for social, communicative interaction. First, adding the knowledge of temporal interaction structure enables natural language interaction on a time scale much longer than current chatterbot technologies. Second, adding support for hierarchical interaction decomposition, where an interaction is represented as a collection of smaller, simpler elements, simplifies the authoring of complex engagement. Third, adding support for the concurrent engagement of these elements enables engagement in interleaved, naturalistic communication. The resulting decomposition supports redundancy of representation, graceful performance degradation through the simultaneous engagement of behaviors on different levels of abstraction, and the stochastic approximation mechanism increases robustness in the face of noise and ambiguity.
Story Games and the OPIATE System, by Chris R. Fairclough, University of Dublin – Trinity College
This thesis presents a new approach to creating game mechanics, utilizing a number of key concepts that result in an interaction scheme that engages a player with a story, while allowing the player the freedom to interact with and alter that story as it happens. A story director agent was developed that uses case-based planning of skeletal plot scripts, modelled on Propp’s morphology, and the dynamic adaptation of these plans. This agent was incorporated into a social simulation engine that a player interacts with through controlling one of the characters therein. The story director and social simulation are symbiotically linked, with a feedback mechanism that ensures plots are planned consistently with the simulation.
July 28th, 2005 at 6:54 pm
Autonomous Expressionism and Network Arts:
New Paradigms in Art, Emotional Interaction, and Information Retrieval “Abstract: In this dissertation, I depart from traditional computer science metrics and methodologies and introduce a new framework for building computer systems with an emphasis …
July 30th, 2005 at 6:58 am
I’m reading through these theses and I’m quite impressed with the new technology, but I also feel a growing unease. I am worried that the technical side of things is outstripping the artistic side. System after system seems to be being built with little evaluation. One or two proof-of-concept works are generated, and then the system is shelved, in favour of the next.
I fear that without having a group of authors attempt real work with any particular engine (over several years), we are never really going to know what is good or bad about a particular technology, and how it could realisticly be improved. I worried that so much effort is spent buidling “the next great thing” which fixes the perceived flaws of what came before, based on scanty evidence of what those flaws are or how they should be fixed.
Meanwhile, the artists in the IF community are truly pushing the limits of their tools (traditional adventure game engines). Has anyone asked them what they would like to have? Do we have anything solid to offer them, so they can beat on it for a few years and turn out a couple of dozen works of art?
Malcolm
July 30th, 2005 at 11:44 am
Good points; a couple of comments. Research efforts such as these are disparate sometimes, but sometimes build upon each other, if not by literally sharing code, at least by influencing each other in approach or conceptualization. Also, I think it’s safe to say that, in the big picture, researchers / artists / game developers are still in the early stages of trying to understand how to apply non-trivial AI to art and entertainment; it makes sense that there will be a wide variety of experiments and approaches. Most researchers / artists / game developers still “roll their own”. But I agree with you, I’d hope over time we’d start to see people building upon each other’s work more. They’re going to need to, to make more substantial progress.
For example, Facade builds directly upon a decade of pioneering work by the Oz Project researcher/artists, in particular the work of Joe Bates, Bryan Loyall, Scott Reilly, Peter Weyrauch, and Brenda Laurel (who consulted on Oz), as well as applying some relatively mature forward-chaining rule technology (in our case, we used Jess); and some of the more powerful features of Java (e.g., reflection); those are big reasons why the project was able to achieve what it did.
> Has anyone asked [the IF community] what they would like to have? Do we have anything solid to offer them, so they can beat on it for a few years and turn out a couple of dozen works of art?
It’s the exception, rather than the rule, for a research group to create a user-friendly authoring system of any kind; it’s a good deal of extra work (and required skill) beyond the research itself to make the technology / research results that accessible. Processing is probably the best recent example of the kind of thing you’re asking for; Alice and AIML are others. … As we’ve mentioned in the past, a component of Facade, the behavior language ABL, will be publicly released in the not-too-distant future, free for academic / non-commercial use, but it needs to be hooked up to a world (animated, text, robot, what have you) to be useful. The initial release will have basic hook-ups to the Unreal Tournament world (which may or may not be an artistically fertile world for artists to build in; they may need to do the work to hook it up to other worlds, such as MUDs, other animation engines, their own robots, etc.).
July 30th, 2005 at 3:30 pm
It’s not just an issue of building on each other’s work. There’s also the issue of doing the research in the context of creating produced experiences. I’ve long argued that in Expressive AI research, the AI research must be done in the context of building specific interactive art and entertainment experiences. Pushing on concrete experiences generates AI research questions that wouldn’t come up otherwise. In fact, you don’t even know if you’re asking interesting research questions and building relevent architectures unless you’re doing it in the context of a concrete experience. The challenge of course is to conceptualize interactive experiences that really require significant steps forward in AI, while not requiring impossible, AI-complete solutions. I’ve seen research work in Expressive AI founder on both sides: either prototype experiences that could have been built by dusting off some finite state machines or simple rule engines, and thus do not push the AI in interesting directions nor provide a compelling case for the more complex solution, or prototype experiences that make very broad claims about extremely hard problems, but where the prototype itself offers an extremely simple solution that falls far short of the broad claims. The latter problem can come up quite easily in interactive story research, because of how hard the problem is.