Inconsistency, Bias, Construction and Abstraction

This paper has argued that the ability to have inconsistent knowledge, to hold decision-making bias, to construct concepts from low to high levels of representa- tion, and to abstract away details may be the key to unlocking more human-like cognitive classi_cation abilities in computer systems. Each of these areas holds very diffcult problems for future research. Technologies that could be derived from these capabilities could have a qualitative impact on how people use com- puter systems and how they affect quality of life.

In particular, some of the techniques that may be possible through these approatechniques are technologies that would allow a human teacher to teach a language to a computer, technologies that can detect types of malware that have never been seen, algorithms to classify data according to its information content rather than through matching text strings or by correlation with tagged content.

Each of these areas also holds other ethical considerations, such as how these technologies should and should not be used. These considerations should be taken into account as research proceeds in these areas.

Bio

Adam Bryant is a computer science and cyber security researcher at the Air Force Research Laboratory interested in autonomic trusted sensing, computational intelligence, formal security modeling, and cognitive science. He served 9 years in the U.S. Air Force as a missile maintenance technician and later as a communications and information officer. He earned a bachelor of science degree in Social Psychology from Park University in 2002 and two master of science degrees from the Air Force Institute of Technology in Information Resource Management and Computer Science in 2007. He is currently enrolled as a Ph.D. student studying Computer Science at the Air Force Institute of Technology. Mr. Bryant also enjoys writing, playing, arranging, and recording music, drawing, and painting.