Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This study explores how different AI interaction modes—Tutor, Collaborator, and Solver—affect cognitive engagement and problem-solving behaviors in high school students, utilizing both behavioral metrics and electroencephalographic (EEG) data. By analyzing 48 participants across two quizzes, the research identifies significant differences in behavioral measures such as Initiation, Processing, and Stress, although EEG results did not achieve statistical significance. The findings suggest that distinct AI interaction modes may influence learning behaviors and highlight the need for further investigation into their neural correlates.
Distinct AI interaction modes can significantly alter high school students' cognitive engagement and problem-solving behavior, revealing nuanced effects that challenge the binary view of AI exposure.
As artificial intelligence (AI) is rapidly integrating into education, concerns have emerged regarding its potential implications on cognitive engagement and problem-solving behavior. However, existing research largely treats AI exposure as a binary condition (AI vs. no-AI), with limited differentiation between interaction modalities and post-exposure effects. This study investigates whether distinct AI interaction modes (Tutor, Collaborator, Solver) influence frontal EEG spectral activity. Electroencephalography (EEG) data and quantified behavioral metrics were recorded from 48 study participants (24 males, 24 females; ages 14-18) across two counterbalanced quizzes in a within-subject design. Statistical analyses included Friedman tests, repeated-measures ANOVA, paired t-tests, and effect size calculations. Behavioral changes were mathematically analyzed in an observation matrix of three characteristics -Initiation, Processing, and Stress-measured on an ordinal scale. Each mode showed significant differences in all three behavioral measures. Descriptive EEG patterns in AI interaction mode were observed, and the possibility of short-term carryover effects of AI was explored. Although the EEG data did not reach statistical significance, the patterns observed across the three AI interaction modes warrant further investigation. This study provides preliminary behavioral evidence and investigative electrophysiological observations, exploring possible AI-interaction-mode-based differences in neural activity and behavior, while establishing a replicable framework for future human-AI interaction studies.