Can AI Counter Degeneration in Large Cooperatives?
Image credit: Alpine region of South Tyrol. Image licensed under Creative Commons Attribution–NonCommercial–ShareAlike (CC BY-NC-SA).
The Participation Challenge in Large Cooperatives
How can Artificial Intelligence leverage joint decision-making and increase democratic participation? Is the usage of AI a facilitator of cooperative economics, or is it only a threat to collective action? Are there organizational barriers which can be overcome by applying Large-Language Models (LLMs) in cooperatives, or do new challenges arise? Are there already use-cases in which AI can help cooperatives to unlock the knowledge of their members and thus lead to a competitive advantage of the cooperative model?
My name is Jeremias Meyer. I am a Ph.D. researcher at WU Vienna University of Business and Economics in Austria. With a background in Industrial Engineering, I am especially interested in the intersection of technology and organizational practices. In my dissertation, I explore the conceptual and organizational dimensions of cooperative governance in light of the use of AI. The following essay outlines the research I will conduct over the coming months, with my final essay and November presentation offering further insight into the findings.
A Multiple-Case Study Approach
As part of the ICDE Fellowship, I am conducting a multiple-case study of five cooperatives in Austria and South Tyrol, Italy, in which I use an organizational intervention to test LLMs in a cooperative context. The cooperatives examined in this study each have more than 100 members and a long-standing tradition of cooperative culture. The members are either small family businesses or self-employed individuals and are therefore primarily linked by a shared economic interest.
From Degeneration to Participation
Although the cooperative is for most of the members their main economic income, the participation is streamlined towards voting in the general assembly, and the managerial board makes most of the decisions. While this large number of members enables these cooperatives to gain economic market power, they face a declining connection towards their members. In the literature on cooperatives, this phenomenon is referred to as degeneration, meaning that cooperatives become increasingly similar to conventional corporations. This can result in low participation in general assemblies and difficulties when board positions need to be filled.
AI-Supported Member Deliberation
Through this organizational intervention, I aim to address this problem by organizing online video conferences in which members are invited to discuss the current challenges and barriers they face in their business environment. In contrast to smaller worker cooperatives, where direct communication among members is a core part of cooperative life, this intervention is fundamentally new for the cooperatives I am investigating. It gives them the relevant space to get to know other cooperative members and address topics, which are currently relevant for their business. This could include, for example, high energy prices or climate change mitigation strategies. Online video conferences theoretically make it possible to bring all members together at relatively low cost for both the cooperatives and their members. During these conferences, break-out rooms are used to discuss with a smaller number of fellows. These sessions are recorded, transcribed using AI, and subsequently used to produce executive summaries with LLMs. The Executive Summaries will provide an overview of the different opinions, possible fields of interaction and strategic development. By using a Retrieval-Augmented Generation System, the transcripts of the video conference become the data source for the used LLM. Thus, it is traceable, on which information the executive summaries are built. Together with the management and the board we elaborate on the topics discussed beforehand. This should encourage the members to join the online meetings, if the topic is important for them. Additionally, I produce short videos for each cooperative to inform members about the project and motivate them to participate.
While LLMs offer new ways to capture and summarize a range of member perspectives, their use also raises organizational questions, especially when combined with efforts to increase member participation. My theoretical perspective therefore examines how cooperative governance may change when more systematic information about members’ needs and views becomes available. Will the board act on this information in the interests of the members? Will members participate in the online conferences at all? Or are they primarily interested in the cooperative as an economic arrangement, with little interest in deeper forms of involvement?
A manager of one of the cooperatives said, “they [the other cooperative members] stop thinking, as soon as they deliver their crops”. Are members really not interested in further participation, or is it just the view of the managers? Or another one said: “If the two most influential members say what we should do, that is usually what happens. But they are also the two most active members.” Might increased participation amplify the power of less active members?
New Institutional Economics and Participation
To address this question from a theoretical perspective, I take a closer look at the effects of this organizational intervention on cooperative governance. My understanding of cooperative governance is grounded in the literature of New Institutional Economics (NIE). NIE can be understood as an umbrella term for a stream of theories which build upon the bounded rationality of actors, information asymmetry among board and members, and reciprocity. NIE has been criticized for being deterministic in its description of organizations, however, I use it only as an analytical lens, rather than a predictive theory. The four main theoretical strands that I use as a framework are transaction cost economics, agency theory, property rights theory, and collective action theory.
The underlying idea behind transaction costs is that members of an organization face cost when they search for new information or take time to make informed decisions. Especially in democratic organizations, the costs for making joint decisions are often highlighted. In theory, the usage of online – video conferences should lower these costs for the coop-members accordingly. Further, Agency theory discusses the relationship between the owners of an organization (the coop members) and the managers and board members acting on their behalf. While managers and board members may have more information about the business environment and may also pursue their own interests, low participation in general assemblies and social events may equally reflect a lack of information about the actual needs and ideas of other members.
Thus, Agency theory helps to investigate the role of managers and board members, when it comes to additional participation and increased availability of information on the member interests. Will the board act accordingly, or will they just follow their own strategy? In the context of my study, property-rights theory is mainly important for the question of information ownership. Who can use the information? Will the information be stored after leaving the organization? Who has access to potential benefits deriving from the discussions? The technical system behind the intervention is based on big AI companies, thus increasing the dependency on outside infrastructure.
While the focus of my work is primarily on the question, how AI can leverage the collective intelligence of cooperative members, it is always important to embed it in the bigger discussion on AI-dependency. Finally, collective action theory is primarily concerned with free-riding in organizations and with how it can be mitigated, potentially through greater social interaction.
The ways of discussing and interacting in a democratic setting have been described in the political science literature and were already previously adapted to the context of cooperatives. Here, the overall impact on the broader society is often highlighted. The so-called spillover thesis holds that when cooperative members learn to negotiate differences and reach compromises within a democratic organizational setting, these practices may have positive effects on the wider society.
The format of members participating in discussions, similar to my setting, have been described from a deliberative democracy perspective. Where the members can engage freely and speak about their needs and ideas in a power free space. But are these online-conferences really powerfree? How to deal with powerful members as previously described “the two big members say what we should do, then they get it accordingly. But I mean, they are also the two most active ones.” Contrary to the deliberative democracy perspective, the agonistic democracy theory describes modes of participation in which contrary opinions can meet and be in conflict with each other. Which is more suitable to describe the settings of my coops?
The outcome of my research is explorative:
It remains an open question whether members are interested in digital formats, whether they will attend the meetings, whether online discussions are an effective way to bring cooperative members together, whether AI-generated summaries will provide useful information, and how the board will respond. At present, the literature offers limited insight into how member participation can be strengthened and how new formats of collective action can be developed.
To investigate these issues, the case-study research is based on mixed-methods. Meaning that I use both, qualitative and quantitative methods to draw a bigger picture. To analyze the outcomes, I draw on field notes taken during calls and discussions with the cooperatives, transcripts of the online discussions, and semi-structured interviews conducted after the organizational intervention. From a quantitative perspective, the members will fill out a survey before the intervention started, and again, half a year after the intervention. The survey is designed to assess levels of trust in the board, fellow members, and management.
Initial Findings on Participation
My preliminary findings, based on field notes from the kick-off events, indicate that the board is genuinely interested in increasing member participation but has little confidence in the capabilities of the members. Since most members live in rural areas and work in non-office jobs, they are not accustomed to digital forms of interaction. Discussions during the kick-off meetings with board members and management showed that all of the cooperatives in this study struggle with member participation, since members often do not attend common events. Some of the members said in a discussion before we started the intervention: “The digital format will change anything – always the same people are coming.” The intervention is scheduled to begin in mid-April 2026, with interviews to follow in January 2027.
In my research, the use of AI is still relatively limited. At this stage, AI is used mainly to transcribe and summarize the diverse views of members. One practical issue we are still working through is how well the system can handle strong Austrian dialects. Further development promises even more possibilities to use AI in online video conferences. For instance, AI could serve as a facilitator in small-group discussions by helping to bridge polarized views and prompting less active members to participate. New forms of collaborative work environments are emerging; however, it remains uncertain how they will unfold under the governance constraints described above.
Toward AI-Supported Cooperative Governance
Through the ICDE Fellowship, I hope to learn more about alternative ways to prevent degeneration and foster greater member participation. Although democratic values are fundamental to cooperatives, member involvement should not be seen as an end in itself, but as a way to strengthen strategic insight and decision-making. Will increased participation lead to a better understanding of the business environment? By orienting towards the member needs, the cooperatives have the potential to even better fulfill their actual purpose – increasing the benefits for the members.
Specifically, I am looking forward to collaborating on projects in which cooperatives already use AI tools in their daily practice to solve the specific challenges which come with cooperative organizing. How was it introduced, in which areas may the AI operate? I would also like to explore the theoretical implications of AI use for governance and modes of organizing in the cooperative context. Here I would like to broaden my horizon to other theoretical positions, such as paradox theory, or other theoretical lenses describing the power and hegemonial relationships among cooperative members.
Lastly, I would like to critically examine the increased dependency on BigTech companies when using AI in coops. While Big Tech provides relatively inexpensive solutions for the initial implementation of AI, it also makes cooperatives dependent on external infrastructures. From the perspective of inter-cooperative solidarity, I am interested in how cooperatives can build an AI ecosystem that serves their needs and advances their common goals.
How can artificial intelligence (AI) support joint decision-making and strengthen democratic participation? Can AI serve as a tool for cooperative economics, or does it mainly threaten collective action? Can large language models (LLMs) help cooperatives overcome organizational barriers, or do they create new ones? Are there already use cases in which AI can help cooperatives to unlock the knowledge of their members and thus lead to a competitive advantage of the cooperative model? For existing cooperatives, AI can be both an opportunity and a threat.
I am thrilled to find out!
About the author:
Jeremias Meyer is a Ph.D. student at the Research Institute for Cooperation and Cooperatives at the Vienna University of Economics and Business