Download to read offline
Uploaded byPeter Brusilovsky
Human Interfaces to Artificial Intelligence in Education
Human Interfaces to Artificial Intelligence in Education discusses: 1) The need for transparency, interactivity, and human-centered design when developing AI systems for education to address issues like lack of ability to inspect AI decisions and lack of trust in AI recommendations. 2) Approaches like explainable AI, visualizing learner models and domain models, and natural communication with AI systems to provide transparency and user control. 3) Examples of open learner model visualizations and explanatory recommendations that make learner knowledge and AI recommendations more transparent.
More Related Content
Similar to Human Interfaces to Artificial Intelligence in Education
Human Interfaces to Artificial Intelligence in Education
- 1. Human Interfaces toArtificial Intelligence in Education Peter Brusilovsky University of Pittsburgh Vasile Rus University of Memphis
- 2. • Long termproblem! – The experience of “expert systems” • Modern issues – Possible biases of AI-based decisions with no ability to inspect – Lack of trust to decisions recommendations coming from AI – Limited ability to control or impact AI 2 Human-AI Communication
- 3. • Transparent AI •Human-Centered design of AI systems • “Natural” communication with AI systems • User interfaces for recommender systems – Controlling and explaining recommendations 3 Growing stream of research
- 4. What’s about AIin Education? 4 A classic view on an AI-Ed system architecture
- 5. What are PossibleSolutions? • Explain – Why a specific learning activity item is good at the current point? • Visualize – What is the current state of the learner model • Communicate – Talk to an AI-Ed system in a natural way • Control – Edit or negotiate your learned morel – Express your goals and preferences in the learning process 5
- 6. Two Sides ofthe Same Coin Explain Visualize CommuicateControl 6 Transparency Interactivity No full transparency without interactivity Control is challenging without transparency
- 7. VISUALIZE! Making domain andlearner model visible and explorable 7
- 8.
- 9. John O'Donovan, BarrySmyth, Brynjar Gretarsson, Svetlin Bostandjiev, and Tobias Höllerer. 2008. PeerChooser: visual interactive recommendation. CHI '08 PeerChooser (O’Donovan, 2008) 9
- 10. Open Learner Models 10 Bull,S., Brusilovsky, P., and Guerra, J. (2018) Which Learning Visualisations to Offer Students? In: V. Pammer-Schindler, M. Pérez-Sanagustín, H. Drachsler, R. Elferink and M. Scheffel (eds.) Proceedings of 13th European Conference on Technology Enhanced Learning, EC-TEL 2018, Leeds, UK, September 3–5, 2018, Springer, pp. 524–530.
- 11. Open Learner Model(ELM-ART) 11 Weber, G. and Brusilovsky, P. (2001) ELM-ART: An adaptive versatile system for Web-based instruction. International Journal of Artificial Intelligence in Education 12 (4), 351-384.
- 12. Mastery Grids: OpenSocial LM 12
- 13.
- 14. EXPLAIN! Make it moreclear for students why specific recommended learning activities are recommended and how they relate to their knowledge and learning goals 14
- 15. Transparent Educational Recommender? 15 Lackof Transparency !
- 16. ATEC Workshop 2 01 9 Los Angeles 16 Explanations with OLM Barria-Pineda,Jordan,andPeterBrusilovsky.2019."Explaining EducationalRecommendationsThroughaConcept-levelKnowledge Visualization."InProceedingsofthe24thInternationalConferenceon IntelligentUserInterfaces:Companion,103--04.NewYork,NY,USA: ACM.
- 17.
- 18. Remedial Recommendations Remedial Visual explanations Relatedconcepts highlighted Knowledge estimates as bar-chart Recent success rate as bar-color Warning sign on “struggled” concepts 18
- 19. Remedial Recommendations Textual explanations #of “struggled” concepts # of “proficient concepts” (Knowledge Est. > .66) 19 Barria-Pineda, Jordan, Kamil Akhuseyinoglu, and Peter Brusilovsky. 2019. "Explaining Need-based Educational Recommendations Using Interactive Open Learner Models." In International Workshop on Transparent Personalization Methods based on Heterogeneous Personal Data, ExHUM at the 27th ACM Conference On User Modelling, Adaptation And Personalization, UMAP '19. Larnaca, Cyprus.
- 20. Activity 1 • Identifyplaces and context where AI could be used to enhance • Each table to suggest 3-7 contexts with votes – how many people at the table would like to talk about this context both to present existing experience and to learn 20
- 21. Activity 2 • Groupcontexts into 3 clusters focused on similar types/context of using AI • Each table focus on a groups of similar types/contexts. • The goal is to exchange information and generate a list of issues (ethical and technical)/controversies/biases/possible problems associated with AI use in each context 21