An Exploration of the Value Alignment Problem in Embedding Multimodal Large Models into Ideological and Political Theory Courses
Abstract
technology and the fundamental objective of value guidance. Thus, value alignment (VA) becomes a key issue. This paper extends the meaning of VA from the field of the ethical aspects of artificial intelligence to this specific case of education. It explains that the adoption of multimodal technologies increases the demand and the difficulty of such an alignment. It also points out three main dilemmas: biases in data sets,
the black-box of the algorithm, the loss of critical reflection with the immersive experience. In order to realize the value alignment, it is necessary to create synergies in a technological, institutional, subjective and conceptual way, in order to make sure that the technology will support
a moral and virtue-based education.
Keywords
Full Text:
PDFReferences
[1] Li, H. R. (2025). Multimodal large models empowering the discursive innovation of ideological and political education. Ideological and
Theoretical Education, (11), 84–90.
[2] Wu, W., & Shen, Y. (2025). Four-dimensional empowerment of multimodal large models for the narrative of ideological and political
education. Journal of Beijing University of Posts and Telecommunications (Social Sciences Edition), 27(5), 73–82.
[3] Sun, W. P., & Liu, H. Y. (2025). Several dilemmas facing value alignment in generative artificial intelligence. Hunan Social Sciences, (4),
1–8.
[4] Sun, W. P., & Liu, H. Y. (2026). From guidance to symbiosis: A philosophical analysis of pathways to value alignment. Philosophical
Analysis, 17(4), 3–17+195.
[5] Lin, B. H., & Xu, D. (2026). The logic and approach of shaping AI values with common human values. Ideological and Theoretical
Education, (9), 55–62.
DOI: http://dx.doi.org/10.70711/neet.v4i10.10287
Refbacks
- There are currently no refbacks.
