Artificial intelligence-mediated informal digital learning of English (AI-IDLE) might strengthen second language (L2) learners' motivational self-concept (e.g., the ideal L2 self) and enhance their foreign language enjoyment (FLE) by enabling them to build confidence, engagement, and willingness to practice their English skills in a self-directed, instant feedback, and non-judgemental learning environment. In our explanatory mixed-method study, we collected questionnaire data from 299 Chinese undergraduate English as a foreign language (EFL) learners and interviewed 12 of them. Structural equation modelling showed that students who participated in AI-IDLE more often reported a clearer ideal L2 self and greater FLE, but those with a greater ideal L2 self did not report more FLE. In addition, gender did not moderate the impact of AI-IDLE on FLE. Analysis of the interview data not only corroborated the quantitative results but also highlighted that while EFL learners can acquire a sense of FLE and vivid ideal L2 selves as they agentively negotiate the affordances of generative AI for informal language learning purposes, the sense of FLE and motivational force may shift across contexts to shape their continued investment in AI-IDLE practices. By comparing and integrating the quantitative and qualitative insights, this study highlights the pedagogical potential of AI-IDLE activities that can strengthen EFL learners' motivation, enjoyment, and commitment to English learning.