Ensuring Pedagogical Value in Very Small Networks: Towards the Network Effects of Generative AI

Date/Time
Date(s) - 10/07/2025
1:30 pm - 3:30 pm

Led by Ian Lumb

York University

Social media involves the use of interactive technologies, a means for creating and sharing content, and the formation of networks and communities. Since the Ethernet-era introduction of Metcalfe’s Law in the 1980s, there has been a predisposition towards the quadratic impact of connected users as the means for establishing value. While the traditional notion of a ‘network effect’ remains valid at the unfathomably large scales of today’s Internet and its social networks, the much more recent introduction of ChatGPT needs to be factored in. ChatGPT is not regarded as social media; it lacks the innate means to facilitate the formation of networks. As a text- or voice-based natural language interface, ChatGPT facilitates interactions between a human (the user) and a Large Language Model (LLM). Yet, the versatility of generative AI clients such as ChatGPT is impressive. For example, with very minimal prompting, GenAI clients allow personas to be established; thus, users can engage naturally and deeply in, for example, a  scientific discourse. Through use of a notebook-style interface, the different perspectives in just a few documents can be surfaced and experienced by exploiting grounded Retrieval Augmented Generation (RAG, a method for improving the quality of responses through use of external data). Thus, it is argued here that there exists pedagogical value in Very Small Networks (VSNs) when the amplifying effects of generative AI are factored in. From GenAI clients to grounded RAG, and more, session participants will be exposed to hands-on strategies and tactics that have been in use across multiple courses for non-scientists for more than a year. Moreover, as agentic AI continues to emerge, prospects for the pedagogical value of VSNs appear all but guaranteed. Finally, because any introduction of AI must necessarily include consideration of responsibility, the session will close with a facilitated discussion on the potential benefits and pitfalls of making AI more of a social media.

Ian Lumb was introduced to regression methods before they were ‘rebranded’ as Machine Learning. Since at least this time, Ian has remained curious about the uptake of computing in support of science. As someone who has been teaching science to non-scientists for some time, Ian’s curiosity continues to be stimulated through the use of AI in the learning process. Using specific examples drawn from his experiences in large, general-education courses, Ian maintains that AI has a lot to offer when it comes to teaching and learning. However, as Ian seeks to encourage group work within these courses, his current interests emphasize sociotechnical gaps – that is, those challenges and opportunities required to enable collaboration involving AIs (aka. agents) within very small networks. In parallel, as a technical consultant, Ian is responsible for the design, implementation, and support of solutions that are typically built with AI.