Improved Performances and Motivation in Intelligent Tutoring Systems: Combining Machine Learning and Learner Choice

Benjamin Clément,Hélène Sauzéon,Didier Roy,Pierre-Yves Oudeyer

2024-01-16

Abstract:Large class sizes pose challenges to personalized learning in schools, which educational technologies, especially intelligent tutoring systems (ITS), aim to address. In this context, the ZPDES algorithm, based on the Learning Progress Hypothesis (LPH) and multi-armed bandit machine learning techniques, sequences exercises that maximize learning progress (LP). This algorithm was previously shown in field studies to boost learning performances for a wider diversity of students compared to a hand-designed curriculum. However, its motivational impact was not assessed. Also, ZPDES did not allow students to express choices. This limitation in agency is at odds with the LPH theory concerned with modeling curiosity-driven learning. We here study how the introduction of such choice possibilities impact both learning efficiency and motivation. The given choice concerns dimensions that are orthogonal to exercise difficulty, acting as a playful feature.

Computers and Society,Artificial Intelligence,Machine Learning

What problem does this paper attempt to address?

### What problems does this paper attempt to solve?

This paper aims to solve the following two main problems:

1. **Motivation assessment in personalized learning**:
- Although the adaptive learning system based on Learning Progress (LP), such as the ZPDES algorithm, has shown advantages in improving students' learning performance, its impact on students' intrinsic motivation has not been fully evaluated. Therefore, this study hopes to fill this gap and evaluate the actual impact of the LP model on students' intrinsic motivation.

2. **The impact of students' choice rights in the learning path**:
- Although the original ZPDES algorithm can recommend the most appropriate exercises for students according to their learning progress, it does not give students the opportunity to choose. However, allowing students to make choices during the learning process may enhance their sense of autonomy and intrinsic motivation. Therefore, this study introduced a new system (ZCO), which combines adaptive recommendation based on learning progress and limited choice rights, to explore the impact of this design on students' learning effectiveness and motivation.

### Research background

- **The need for personalized learning**: As the class size expands, it is difficult for teachers to conduct personalized teaching according to the characteristics of each student. Educational technology (such as Intelligent Tutoring System, ITS) can provide personalized learning paths in an automated way to help teachers and students.

- **Learning Progress model (LP model)**: This model emphasizes the virtuous cycle between curiosity - driven learning and learning progress. By maximizing learning progress, students' intrinsic motivation can be stimulated, thereby improving learning effectiveness.

- **ZPDES algorithm**: This is an intelligent tutoring system based on the multi - armed bandit algorithm, which can dynamically adjust the recommended exercise content according to students' learning progress. Although ZPDES performs well in improving learning performance, it lacks the assessment of students' motivation and does not allow students to make choices in the learning path.

### Research methods

- **Experimental design**: The researchers conducted a randomized controlled trial (RCT) involving 265 children aged 7 - 8 from 24 classes in 11 schools. They used the Kidlearn ITS platform to compare the learning effectiveness and motivation under four conditions:
- **Predef**: Pre - defined linear sequence, no choice right.
- **PCO**: Pre - defined linear sequence, with choice right.
- **ZPDES**: Adaptive recommendation based on learning progress, no choice right.
- **ZCO**: Adaptive recommendation based on learning progress, with choice right.

### Main findings

1. **ZPDES is more effective than the pre - defined sequence**:
- Under the ZPDES condition, students' learning performance has been significantly improved, especially with higher scores in the math test. This verifies the effectiveness of the personalized learning path based on learning progress.

2. **The impact of choice rights on learning effectiveness is context - dependent**:
- Under the ZPDES condition, giving students choice rights further improves learning effectiveness and intrinsic motivation; while under the pre - defined sequence condition, choice rights instead reduce learning effectiveness. This indicates that choice rights can play a positive role only in the personalized learning path.

3. **Emotional experience and motivation**:
- Allowing students to choose exercise objects can significantly enhance their emotional experience and intrinsic motivation, especially under the ZCO condition.

### Conclusion

This study shows that the personalized learning path based on learning progress can not only improve students' learning performance but also enhance their intrinsic motivation. In addition, giving students appropriate choice rights can further enhance these effects, but this choice right must be combined with an effective personalized learning path to have a positive impact.