COGS 6240 - Logic and Artificial Intelligence (Spring 2006)
Main Page
Instructor: Dr. Selmer Bringsjord (selmer@rpi.edu)

Place: Carnegie 201
Time: Mondays and Thursdays, 12:00 PM - 1:50 PM
Course Description

This course is offered yearly in the Spring; each time the topic changes, but each time the general orientation falls under the intersection of logic and artificial intelligence (AI). Three attributes characterize the general approach taken in the course, and in the R&D that underlies and drives it: “Strong-Weak,” “Neat,” and “Logicist” (= “Logic-based”). In short, this means that (i) our focus is on engineering (and laying a foundation for engineering) artificial agents capable of person-level intelligent behavior, not on such glassy-eyed notions as literally building conscious persons (Strong-Weak AI); (ii) I believe that progress in AI will only be made if the relevant work is firmly grounded in a precise mathematical framework (Neat AI); and (iii) a prominent part of my mathematical framework is formal logic, which includes not just elementary logic (e.g., first-order logic) of the sort that introductory AI textbooks are fond of, but much more expressive logics, e.g., those up to the task of representing and reasoning about knowledge (Logicist or Logic-Based AI), and those involving non-deductive (e.g., abductive and inductive) reasoning.

Logic-based AI in connection with intelligence analysis (IA) will drive the Spring 06 version of this course. Over the last 2+ years, I have had the privilege of learning about IA through interacting with intelligence analysts, reading reports they have produced, and reading papers and books about IA. This education has happened in connection with research and development I’ve carried out (with a number of colleagues) in ARDA’s NIMD program, and with its IKRIS program. This R&D has been devoted, at least in large part, to designing, building, and refining the Slate system. Lead developers of the Slate system are Andrew Shilliday and Joshua Taylor. Slate is an intelligent “superagent” designed to assist human reasoners, including intelligence analysts, and those studying intelligence analysis.