Grand Challenges for Intelligent Tutoring Systems in STEM: Progress and Perspectives
  • Location: Special Event at the Intelligent Tutoring Systems 2014 Conference
  • Time: Sunday, June 8, 2014
    • Presentations:            1:05PM - 2:15PM (Garden Lanai Room)
    • Panel & Discussion:    6:25PM - 7:10PM (Plumeria)
  • Organizers: The University of Memphis (Xiangen Hu, Benjamin Nye, Art Graesser) in coordination with the University of Massachusetts (Beverly Woolf), Worcester Polytechnic Institute (Neil Heffernan), Arizona State University (Kurt VanLehn), and Raytheon BBN Technologies (Bruce Roberts)

Home | Interest | Format | Outcomes 

All attendees of the 2014 Intelligent Tutoring Systems Conference are encouraged to attend this exciting event.  This special event has investment from eight prominent ITS research groups, two research program representatives, and an industry expert. We believe that the larger ITS community will be interested in hearing about the progress on these large-scale efforts and also appreciate an open dialog on the key challenges for ITS.  The event has two parts: 1) A set of presentations during the lunch session of the ITS conference on June 8 (1:05 Garden Lanai Room) and 2) A panel discussion and open forum later in the day (6:25PM Plumeria Room).  We welcome everyone at ITS 2014 to attend both sessions.

Perspectives
As a sampling of the topics that will be covered, each of the participants has offered a quote about their vision of the Grand Challenges for ITS now and in the future:

  Xiangen Hu (U of Memphis)

Theoretically, identify a unified framework for existing ITS implementations. Such framework should be based on cognitive psychology of learning. Technologically, establish a systematical methods that integrate ITS in both structured learning environments (formal learning) or non-structure learning environments (informal learning). Practically, establish guideline for best practice for scale-up such that integrating ITS with learning (informal or formal) shows significant learning gains.
 Beverly Park Woolf (UMass)



I suggest that in 5 years the ITS community will be able to provide gifted “teachers” and real-time responses for selected learners based on their learning needs and on cognitive, metacognitive and affective knowledge about them. Additional long-term stretch goals include: providing gifted teachers for every learner on the planet; lifelong and lifewide learning for every citizen; universal access to global education; and access to education at a level that is only enjoyed today by the most privileged learners living in stable and peaceful countries. 
 Neil Heffernan (WPI)

I think the most important challenge for intelligent tutoring systems is 1) figure out how to design a system that allows the masses of teachers to contribute questions and feedback (video hints for instance). How do we pull of what Wikipedia has done: get thousand of good teachers involved in giving feedback. In education, we can run randomized controlled trials to learn what pieces of feedback are effective. 
 Kurt VanLehn (ASU)

Pivotal grand challenge: For the field of Intelligent Tutoring Systems, a grand challenge is understanding why ITS have not been incorporated into schools. This challenge is part of a growing curiosity about why excellent research-based instructional activities are so seldom adopted. Indeed, the USA’s Institute of Educational Science is requesting proposals for a $10M center to study this issue. http://ies.ed.gov/funding/ncer_rfas/randd_knowledge.asp.
Bruce Roberts (BBN) 

A major challenge for the ITS community is to provide industrial-strength platforms for creating and deploying effective tutoring systems. These platforms must include easy-to-use authoring tools, an authoring process that ensures high-quality content and automation support for key authoring tasks; e.g., tutor modeling. Barriers to access and scalability can be addressed by offering tutoring as a hosted service readily accessible from any modern browser.
Diego Zapata-Rivera (ETS)

The most important challenges for ITS now and in 5-10 years include: (a) reducing development and evaluation costs, (b) improving scalability of ITS systems and (c) focusing on improving integration into mainstream K-12 education. By using authoring tools and implementing automated testing, it is possible to reduce costs and improve scalability. Producing relevant information about students' knowledge, skills and abilities for teachers, students, parents and administrators can facilitate the integration into mainstream K-12 education.
 James Lester (NCSU)

Intelligent tutoring systems R&D is now at a crossroads. Our community has demonstrated the learning effectiveness of these technologies for a broad range of curricula and student populations, and it is now time to address three central challenges: 1) Creating authoring frameworks to enable scalable development and deployment; 2) Leveraging learning analytics and big data to amplify learning impact at scale; and 3) Fundamentally expanding the capabilities of ITSs to effectively address affect, metacognition, and engagement.
 Heather Holden (ARL)

The most important challenge of current intelligent tutoring systems pertains to the accurate adaptability of instruction for individual learners. It is a two-fold dilemma: What elements of the learner are most necessary to model and how does the system provide the best instructional strategies dynamically as needed based on those elements? Overcoming the current challenge will significantly impact the future of intelligent tutoring systems. The future of intelligent tutoring systems will scale to support small team tutoring. If the current issue is not addressed, the challenge mentioned above will increase exponentially in the future.
Ido Roll (UBC)
 
Intelligent Tutoring Systems (ITS) have achieved impressive success in a challenging, yet well-defined scenario: individual students learning disciplinary knowledge in the classroom. To shape the future of education in the current dynamic environment, our community should break traditional barriers and offer adaptive support for collaboration, focus on self-regulation, motivation, and dispositions, support learning across disciplines, and branch into informal learning environments. To achieve that we should also expand our arsenal of research methods and give more weight to case studies, ethnographies, and other methods that collect rich data and help us develop vision for personalized learning in diverse contexts.
 H. Chad Lane (USC)
https://sites.google.com/site/its2014specialevent/_/rsrc/1399918529504/home/participants/presenters-1/hchadlane-universityofsoutherncalifornia/27.jpg?width=100

I think we should expand our reach and aim for more impressive outcomes. We need to think more about very young learners (0-pre-K), and stop ignoring older learners. We need to lose our information processing baggage and finally close the loop on affect by building systems that not only detect and respond to emotions, but induce specific ones known to enhance learning. We need to look for new frames for understanding learning, like self-determination theory, and consider learning as a behavior change problem. We need to do a lot of stuff.