Students' interactions with an artificial intelligence assistant in a remote chemistry laboratory

Abstract

The integration of artificial intelligence (AI) in education has demonstrated potential for solving individualised learning challenges, particularly through virtual assistants in natural science subjects. This study analyses students‘ interactions with an intelligent assistant integrated into the acid-base titration II remote laboratory to characterise learning needs and identify assigned roles during experimental activities. The intelligent assistant utilised OpenAI GPT-4o, customised with laboratory-specific information. The analysis of interactions was carried out using a mixed methodology combining content analysis with descriptive statistical analysis. The interactions were systematically categorised, revealing four main dimensions in the use of the assistant which are development of remote experiences, data processing, application and conceptual understanding, and guidance for the preparation of laboratory reports. Students positioned the assistant mainly as a tutor for procedural and calculation support, a conceptual support resource connecting experimental practise with theoretical understanding, and a mediator in scientific communication for report elaboration. This role diversification responded directly to students' main difficulties in experimental procedures, concentration calculations, theoretical understanding, and scientific writing. The study validates the necessity of AI-powered pedagogical support, demonstrating the versatility of artificial intelligence in remote laboratories.