Evaluation of the Generalized Intelligent Framework for Tutoring (GIFT) from a Researcher's or Analyst's Perspective.

摘要

Current US Army standards for training and education are group instruction and classroom training, also known as one-to-many instruction. Recently, the Army has placed significant emphasis on self-regulated learning methods to augment institutional training. Per the Army Learning Model, Soldiers will be largely responsible for their own learning. One-to-one human tutoring has been shown to be significantly more effective than one-to-many instruction but it is not practical to assign each Soldier a personal mentor. An alternative to one-to-one human tutoring is one-to-one computer-moderated tutoring using artificially Intelligent Tutoring Systems (ITSs), which have been shown to be effective in promoting individual learning in static, simple, well-defined domains (e.g., mathematics). To be practical, high authoring costs and limited adaptiveness barriers must be addressed. This report describes the outcomes of an evaluation conducted at the US Military Academy to determine initial usability of the Generalized Intelligent Framework for Tutoring, a tutoring architecture constructed with the goal to reduce time and skill needed to construct ITSs while increasing their adaptiveness or ability to act autonomously to optimize user learning. Participating cadets were assigned tasks related to the researcher s or analyst s perspective as part of a course assignment in PL488E, an engineering colloquium. Their thoughts are shared herein along with technical challenges identified by the US Army Research Laboratory based on cadet observations.