Intelligent tutoring systems (ITSs) are computer software systems that seek to mimic the methods and dialog of natural human tutors, to generate instructional interactions in real time and on demand--as required by individual students. Implementations of ITSs incorporate computational mechanisms and knowledge representations in the fields of artificial intelligence, computational linguistics, and cognitive science.

ITS systems create three types of models: Student (e.g., what she knows, what she’s done, how she learns, …), Subject matter (i.e., explicit knowledge as would be expressed by a Subject Matter Expert), and Pedagogical (e.g., how do you teach, in what order, typical mistakes and remediation, typical questions a student might ask, hints one might offer a student who is stuck).

ITS functionality includes everything that a human teacher might do: select (or generate) appropriate material, "set up" the exercise, monitor student activity, give hints during exercises and feedback afterwards, understand why students make mistakes, customize presentation style to the student’s style, ask and answer questions.

Learning and teaching with human tutors and ITSs are characterized by mixed-initiative dialog and, often, the use of natural language. To achieve such interactions, ITSs embody the mechanisms of human tutors with advanced algorithms and computational architectures, such as production rules systems, generative grammars, Bayes networks, hidden Markov models, neural networks, higher-order semantic spaces, fuzzy control systems, and non-linear dynamical systems. With these methods ITSs use tutorial dialogue in natural language to adaptively respond to the learner's emotions and motivational states, and efficiently bridge the man-machine interface, adapting to a learner's multidimensional profile.

In training based on simulation, game and virtual world technology, where students learn from experiences in a virtual training environment, ITS capability is especially important because it is often hard for human instructors to monitor what the student is doing without interfering with her experience. Extending the instructor’s reach is especially important in multiplayer training environments, where no one can keep track of each individual student and team, and in anytime / anywhere learning, where instructors may not be available.

The primary developmental goals of the ITSs community are aligned with ADL's long-term vision: To generate, assemble, and sequence content that dynamically adapts to the learner to optimize learning. Specifically, ADL is actively engaging in research and implementation of the digital knowledge environment of the future in the areas of standards and authoring tools that give instructors the ability to create ITS functionality within a virtual training environment.