Providing discussion summaries and examples has long been recognised as an effective way to enhance learners’ online learning activities. Previous research has primarily focused on human-written summaries and examples as well as their impact on students’ behaviours. However, recent developments in large language models have demonstrated the potential for generating accurate discussion summaries and example posts at human levels. Despite this, limited research has explored the educational impacts of such AI-generated learning support and how it might affect the social capital students receive from their peers. This paper proposes an AI-generated summary-driven learning design (ASLD) to facilitate peer interactions in online discussion forums. A total of 128 students participated in a design-based research study to examine the influence of the ASLD on students’ viewing breadth, temporal changes in viewing activities, and the social capital students gained over 8 weeks. The results indicate that ASLD significantly improves viewing breadth, as well as the overall connectedness of the cohort and individuals’ bridging social capital. However, ASLD did not prevent the decrease in viewing activities towards the end of the intervention. The implications of the findings, along with the future directions for research of AI-generated support in online learning environments, are discussed.