大模型赋能思想政治教育话语的效度、限度与向度

Validity, Limitations and Dimensions of Empowering Ideological and Political Education Discourse with Large Models

摘要: 大模型作为一种基于海量数据训练和复杂结构的机器学习模型，给思想政治教育话语带来了崭新的发展机会。它凭借Transformer架构、跨模态处理技术和高仿真模拟技术，促进了话语内容的优质化、话语形式的沉浸化、话语传播的场景化。但是，也存在着技术偏见消解话语价值，数据质量参差降低话语信度，智能依赖弱化话语主体能力等限度。需要从强化价值引领、开发垂直大模型、提升智能素养三个向度出发，主动创造有利条件，最大程度地发挥大模型对思想政治教育话语的正向赋能作用。

Abstract: As a machine learning model trained on massive data with sophisticated architectures, large models have brought brand-new development opportunities to ideological and political education discourse. By virtue of the Transformer architecture, cross-modal processing technology and high-fidelity simulation technology, they have facilitated the optimization of discourse content, the immersion of discourse forms, and the contextualization of discourse dissemination. Nevertheless, such limitations also exist as technical bias dilutes discourse value, uneven data quality undermines discourse reliability, and overreliance on intelligence weakens the competence of discourse subjects. To maximize the positive empowering effect of large models on ideological and political education discourse, efforts should be made in three dimensions: strengthening value guidance, developing vertical large models, and enhancing digital intelligence literacy, so as to proactively create favorable conditions.