This article summarizes a workshop entitled "Detecting, Repairing, and Preventing Human-Machine Miscommunication," held on 4 August 1996 in Portland, Oregon. The author presents the significant issues raised during the four specific workshop sessions. Research related to achieving robust interaction is an important subarea in AI. Early work concerned the correction of spelling or grammatical errors in a user's utterance so that the system could more easily match them against a fixed linguistic model; work has also been done in the area of speech recognition, attempting to find the best fit of a sound signal to legal sequences of linguistic objects. All these approaches have assumed that the system's model is always correct.

Of all computational systems, robots are unique in their ability to afford embodied interaction using the wider range of human communicative cues. Research on human communication provides strong evidence that embodied cues, when used effectively, elicit social, cognitive, and task outcomes such as improved learning, rapport, motivation, persuasion, and collaborative task performance. While this connection between embodied cues and key outcomes provides a unique opportunity for design, taking advantage of it requires a deeper understanding of how robots might use these cues effectively and the limitations in the extent to which they might achieve such outcomes through embodied interaction. This article aims to underline this opportunity by providing an overview of key embodied cues and outcomes in human communication and describing a research program that explores how robots might generate high-level social, cognitive, and task outcomes such as learning, rapport, and persuasion using embodied cues such as verbal, vocal, and nonverbal cues. Such representations vary from physical artifacts (such as tangible interfaces) to biological forms (such as humanlike agents and robots) and offer templates for understanding and interacting with complex computational systems (Ullmer and Ishii 2000, Cassell 2001, Breazeal 2003).

On October 8-10, 2009, an interdisciplinary group met in Beverley, Massachusetts, to evaluate the state of the art in the computational modeling of narrative. Three important findings emerged: (1) current work in computational modeling is described by three different levels of representation; (2) there is a paucity of studies at the highest, most abstract level aimed at inferring the meaning or message of the narrative; and (3) there is a need to establish a standard data bank of annotated narratives, analogous to the Penn Treebank. We use them to entertain, communicate, convince, and explain. One workshop participant noted that "as far as I know, every society in the world has stories, which suggests they have a psychological basis, that stories do something for you." To truly understand and explain human intelligence, reasoning, and beliefs, we need to understand why narrative is universal and explain the function it serves. Computational modeling is a natural method for investigating narrative. As a complex cognitive phenomenon, narrative touches on many areas that have traditionally been of interest to artificial intelligence researchers: its different facets draw on our capacities for natural language understanding and generation, commonsense reasoning, analogical reasoning, planning, physical perception (through imagination), and social cognition. Successful modeling will undoubtedly require researchers from these many perspectives and more, using a multitude of different techniques from the AI toolkit, ranging from, for example, detailed symbolic knowledge representation to largescale statistical analyses. The relevance of AI to narrative, and vice versa, is compelling.

This editorial introduction presents an overview of the robotic resources available to AI educators and provides context for the articles in this special issue. We set the stage by addressing the tradeoffs among a number of established and emerging hardware and software platforms, curricular topics, and robot contests used to motivate and teach undergraduate AI. Yet it is only recently that physically embodied agents have become a viable tool in the undergraduate AI classroom. Examples of the flurry of activity in this area include competitions and exhibitions, the growing options for lowcost robot hardware and software, and a number of recent workshops and symposia. This special issue of AI Magazine grew out of the 2004 AAAI spring symposium on Accessible, Hands-on AI and Robotics Education.

While online handwriting recognition is an area of longstanding and ongoing research, the recent emergence of portable, pen-based computers has focused urgent attention on usable, practical solutions. We discuss a combination and improvement of classical methods to produce robust recognition of hand-printed English text for a recognizer shipping in new models of Apple Computer's The ANN character classifier required some innovative training techniques to perform its task well. The dictionaries required large word lists, a regular expression grammar (to describe special constructs such as date, time, and telephone numbers), and a means of combining all these dictionaries into a comprehensive language model. In addition, well-balanced prior probabilities had to be determined for in-dictionary and out-of-dictionary writing. Together with a maximum-likelihood search engine, these elements form the basis of the so-called "Print Recognizer," which was first shipped in NEWTON OS 2.0-based MES-SAGEPAD 120 units in December 1995 and has There is ample prior work in combining low-level classifiers with dynamic time warping, hidden Markov models, Viterbi algorithms, and other search strategies to provide integrated segmentation and recognition for writing (Tappert, Suen, and Wakahara 1990) and speech (Renals et al. 1992).

My observations are based on learning sciences research of the past several decades, the possibilities of new technologies of the past few years, and my experience as program officer for the National Science Foundation's Cyberlearning and Future Learning Technologies program. My thesis is that new technologies have potential to transform possibilities for fostering learning in both formal and informal learning environments by making it possible and manageable for learners to engage in the kinds of project work that professionals engage in and learn important content, skills, practices, habits, and dispositions from those experiences. The expertise of AI researchers and practitioners is critical to that vision, but it will require teaming up with others -- for example, technology imagineers, educators, and learning scientists. The articles report on the newest in intelligent tutoring systems and resources (Bredeweg et al. 2013, Rus et al. 2013; Chaudhri et al. 2013), virtual humans and conversational agents (Swartout et al. 2013), assessment and student modeling for personalization (Conati and Kardan 2013, Koedinger et al. 2013), and intelligently controlled virtual environments (Lester et al. 2013). The final article in the set (Woolf et al. 2013), of which I am a coauthor, suggests needs and challenges facing STEM education (science, technology, engineering, and mathematics) that artificial intelligence might address -- mentors for every learner, fostering learning of 21st-century skills, automating assessment in ways that support learning, universal access, and lifelong and life-wide learning.

Alan Turing's decades-old question still influences artificial intelligence because of the simple test he proposed in his article in Mind. In this article, AI Magazine collects presentations about the first round of the classic Turing Test of machine intelligence, held November 8, 1991 at The Computer Museum, Boston. Robert Epstein, Director Emeritus, Cambridge Center for Behavioral Studies, and an adjunct professor of psychology, Boston University, University of Massachusetts (Amherst), and University of California (San Diego) summarizes some of the difficult issues during the planning of this first real-time competition, and describes the event. Presented in tandem with Dr. Epstein's article is the actual transcript of session that won the Loebner Prize Competition--Joseph Weintraub's computer program PC Therapist. In 1985 an old friend, Hugh Loebner, told me excitedly that the Turing Test should be made into an annual contest.

The articles in this special issue of AI Magazine include those that propose specific tests and those that look at the challenges inherent in building robust, valid, and reliable tests for advancing the state of the art in AI. To people outside the field, the test -- which hinges on the ability of machines to fool people into thinking that they (the machines) are people -- is practically synonymous with the quest to create machine intelligence. Within the field, the test is widely recognized as a pioneering landmark, but also is now seen as a distraction, designed over half a century ago, and too crude to really measure intelligence. Intelligence is, after all, a multidimensional variable, and no one test could possibly ever be definitive truly to measure it. Moreover, the original test, at least in its standard implementations, has turned out to be highly gameable, arguably an exercise in deception rather than a true measure of anything especially correlated with intelligence.

However, existing online learning techniques typically applied to robot learning (for example, Hexmoor, Meeden, and Murphy [1997]) differ fundamentally from human learning. Online root learning using robot sensors is not equivalent to autonomous mental development in robots, nor should mental develop-This article describes a workshop on mental development and learning issues that are relevant to both machine and human sciences. It was jointly funded by the National Science Foundation and the Defense Advanced Research Projects Agency and held at Michigan State University on 5 to 7 April 2000. Such systems already exist (for example, systems that use neural network techniques). There is a need, therefore, for increased studies in computational autonomous mental development (CAMD) that are of interest to both machine and human intelligence researchers.

In this article, we describe a deployed educational technology application: the Criterion Online Essay Evaluation Service, a web-based system that provides automated scoring and evaluation of student essays. Criterion has two complementary applications: (1) Critique Writing Analysis Tools, a suite of programs that detect errors in grammar, usage, and mechanics, that identify discourse elements in the essay, and that recognize potentially undesirable elements of style, and (2) e-rater version 2.0, an automated essay scoring system. Critique and e-rater provide students with feedback that is specific to their writing in order to help them improve their writing skills and is intended to be used under the instruction of a classroom teacher. Both applications employ natural language processing and machine learning techniques. All of these capabilities outperform baseline algorithms, and some of the tools agree with human judges in their evaluations as often as two judges agree with each other. Unfortunately, this puts an enormous load on the classroom teacher, who is faced with reading and providing feedback for perhaps 30 essays or more every time a topic is assigned. As a result, teachers are not able to give writing assignments as often as they would wish. With this in mind, researchers have sought to develop applications that automate essay scoring and evaluation. Work in automated essay scoring began in the early 1960s and has been extremely productive (Page 1966; Burstein et al. 1998; Foltz, Kintsch, and Landauer 1998; Larkey 1998; Rudner 2002; Elliott 2003). Detailed descriptions of most of these systems appear in Shermis and Burstein (2003). Pioneering work in the related area of automated feedback was initiated in the 1980s with the Writer's Workbench (MacDonald et al. 1982). The Criterion Online Essay Evaluation Service combines automated essay scoring and diagnostic feedback. The feedback is specific to the student's essay and is based on the kinds of evaluations that teachers typically provide when grading a student's writing. Criterion is intended to be an aid, not a replacement, for classroom instruction. Its purpose is to ease the instructor's load, thereby enabling the instructor to give students more practice writing essays. Criterion contains two complementary applications that are based on natural language processing (NLP) methods. Critique is an application that is comprised of a suite of programs that evaluate and provide feedback for errors in grammar, usage, and mechanics, that identify the essay's discourse structure, and that recognize potentially undesirable stylistic features. The companion scoring application, e-rater version 2.0, extracts linguistically-based features from an essay and uses a statistical model of how these features are related to overall writing quality to assign a holistic score to the essay. Figure 1 shows Criterion's interface for submit-