Saturday night a friend broke off the existing conversation mid-topic and looked me straight on, looking terribly serious. "What do you do?" he asked.
"I'm a Human Factors Engineer for an artificial intelligence software research and design company," I said.
"Oh. So you're, like, smart and shit."
"That's what they tell me."
Just a minute ago, andrh1a IM-ed me:
: I'm not entirely clear on what your job is
: So I amuse myself thinking that you design nuclear wessels.
: I suppose that's all kinds of wildly inaccurate :)
Uh, yes, I admitted. So what with one thing and another, I thought maybe I should write this up.
What I do (for a living)
I work for a small company called Soar Technology. Soartech, as we call ourselves internally, does human behavior modeling and cognitive modeling, both in a rule-based reasoning cognitive modeling language called Soar and also using other languages, some of which we are developing ourselves. We do research and software design for a number of different domains, mostly for some branch of the federal government or military. Some of our main areas of focus are Intelligence Analysis, Robotic Command and Control, and Training (for various purposes). All of these things we investigate and contribute to on projects by ourselves or partnered with one or more other companies and/or universities. Most of our systems are designed to work in simulation rather than in the real world, but we're working on changing that while also retaining/leveraging our simulation-based training expertise.
What I do is I analyze the cognitive work domains that we are planning to support with the systems we're building. At the very least, I need to understand the cognitive work involved - what are the goals and the boundaries of the work we're talking about? What are the rules and procedures? What decisions are involved and what information is needed to support those decisions? To do this analysis I read documentation and talk to or observe Subject Matter Experts (SMEs) and, sometimes, users. One step forward from that, I express what I learn as a set of Requirements for the software designers and scientists to choose to address. Then, depending on what is selected for our focus, I help design displays and other Human-Computer Interaction (HCI) components (HCI is a term used to capture the fact that we're designing dynamic interaction, not just static displays). I also sometimes create formalized task models that are used in a few different ways for our work (see GOMS for instance), including constructing systems that try to identify what the user is up to and change the interface to support that work (what are modestly called Intelligent User Interfaces, or IUI). Finally, I help do user studies - experiments in which we have people use systems and then do qualitative and/or quantitative analysis of their experience and performance.
Since I was hired, my main project has been one called the Intelligent Control Framework (ICF). Working with the U.S. Army Tank Automotive Research, Development, and Engineering Center, we've been investigating how to architecture intelligent agent support of single-user command and control of a heterogeneous group of multiple unmanned vehicles (air and ground, with various capabilities). Using the Future Combat Systems program as a guide for autonomous capabilities we can safely assume, we've explored questions of deontic authority, adjustable autonomy, human interaction with autonomous systems, and robotic team building and coordination. Right now the scientific and engineering world is still working hard on getting to where unmanned vehicles can navigate and perceive the world autonomously. We're trying to ask some of the "What then?" questions of how we might use such unmanned systems intelligently, and we've built a simulation-based testbed of tools to help us do that.
I've also worked on other projects, doing socio-cultural modeling and scenario creation (One month I was assigned to study Nigeria. It was cool. Side note: that is one messed-up and fascinating country.) and, just recently, using storytelling for training purposes --I'm designing a review/event playback interface that makes the actions and reasoning of an interactive training Director agent visible to the user. (see this year's soar workshop update presentation on ISAT for more info). In-house I also manage the team that maintains the corporate website, and act as the user-facing support liaison for the wiki software we use internally, Socialtext. Since Soar builds tools for our own development, two of the more interesting projects I've contributed to have been on the design of Eclipse plugins for Soar and eGLEAN development. Because we are a research company, I also help write research proposals and papers.
Sometimes they send me to cool conferences. In the last couple of years I've attended ACM's Computer-Human Interaction (CHI) and Human-Robot Interaction (HRI) conferences, the Human Factors and Ergonomics Society annual meeting, The Intelligent User Interfaces conference, and symposia on Augmented Cognition, Intelligent Vehicle Systems, Intelligent Vehicle Technology Transfer, the Human Factors of UAVs, and Command and Control Research and Technology. Last year I attended the Army Science Conference, which had a number of presentations by Nobel prize winners and also provided me with a fun opportunity to learn to operate an iRobot Packbot.