Teaching without learning: Is it OK With weak AI?

Kristian Månsson, Magnus Haake

Research output: Chapter in Book/Report/Conference proceedingPaper in conference proceedingpeer-review

Abstract

Two different learning models for a teachable agent were tested with respect to perceived intelligence, the protégé effect, and learning in Swedish grade 5 and 6 students. A strong positive correlation was found between perceived intelligence and the protégé effect, but no significant differences were found between the two different implementations of the learning algorithm. The results suggest that while the perceived intelligence of the agent relates to the induced protégé effect, this perceived intelligence did not correspond to the implemented learning model. This, in turn, suggest that a simple learning model can be sufficient for a teachable agent system, but more research is needed.

Original languageEnglish
Title of host publicationArtificial Intelligence in Education - 19th International Conference, AIED 2018, Proceedings
PublisherSpringer
Pages241-245
Number of pages5
ISBN (Print)9783319938455
DOIs
Publication statusPublished - 2018 Jan 1
Event19th International Conference on Artificial Intelligence in Education, AIED 2018 - London, United Kingdom
Duration: 2018 Jun 272018 Jun 30

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10948 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference19th International Conference on Artificial Intelligence in Education, AIED 2018
Country/TerritoryUnited Kingdom
CityLondon
Period2018/06/272018/06/30

Subject classification (UKÄ)

  • Human Computer Interaction

Free keywords

  • Learning model
  • Learning outcome
  • Perceived intelligence
  • Protégé effect
  • Teachable agent

Fingerprint

Dive into the research topics of 'Teaching without learning: Is it OK With weak AI?'. Together they form a unique fingerprint.

Cite this