SOFTWARE ARCHITECTURE AND PEDAGOGICAL PRINCIPLES OF APPLYING MULTIMODAL ARTIFICIAL INTELLIGENCE TECHNOLOGIES IN THE TEACHING AND LEARNING PROCESS

The paper substantiates the software and pedagogical foundations of introducing multimodal artificial intelligence — models that jointly process text, speech, images and video — into the educational process. A four-layer integration architecture is proposed, and the didactic principles that make such systems instructionally effective are derived from the cognitive theory of multimedia learning and the UNESCO AI competency framework for teachers. Risks of hallucination, bias, data protection and skill erosion are analysed.