I'm often asked what "artificial intelligence" is, but sometimes it is easier to explain what it is not. Generally speaking, AI researchers are NOT building Terminator robots, or Hal 9000, or KITT, or the Enterprise computer. Artificial Intelligence is a broad class of techniques that we can use to make otherwise dumb computers act a little bit smarter in very particular ways. The type of AI that I tend to work with is the kind that learns automatically from data. The wonderful thing about this is that the computer can learn to do something better than I can personally do it; I don't need to speak Russian to teach a computer Russian, I only need to know how to teach the computer to learn from examples of English and Russian text.

The types of AI that I work with are usually called "machine learning" and "data mining". Algorithms in these categories can be applied to a wide variety of different tasks. Some of the domains I've worked with include:

• Social Media analysis & mining (including prediction of latent demographic variables on Twitter, eg gender)
• Topic Modeling, Summarization & Visualization of forum, blog, and social media comments
• Machine Translation (Domain adaptation for automated translation between human languages)
• Automatic News Monitoring (to detect hidden signs of infectious disease outbreak, including SARS)
  
• Market-based robot planning (auction systems for task allocation among groups of robots)
• Network Security (malicious insider threat, intelligent alert filtering)
• Chat and Speech analysis (automated detection of particular communication patterns)
• Information Extraction (identifying events, names, and dates from e-mail)
• Intelligent Tutoring (computer systems that learn what a student is learning)
• Optical character recognition
• Game playing and game theory