Session 5. 6.

Artificial intelligence (AI) encompasses a broad set of technologies with intelligent capabilities, such as learning, image and language processing, planning and possibly reasoning. Depending on its context, AI can function as a club, private, or public good. As a knowledge resource (Ostrom and Hess, 2007), AI’s production, sharing, and use benefit the broader public but, similar to some common pool resources, remain vulnerable to excludability and enclosure. These challenges are especially pronounced in the developing world, where colonial and postcolonial legacies exacerbate these problems. As initiatives around AI and its governance continue to expand, understanding the current state of public involvement in the production, deployment, regulation, and use of AI—and exploring how it can be broadened to promote sustainable development and a more equitable global society—becomes increasingly important.

This paper addresses this question by analyzing the landscape of AI and development through the lens of two frameworks: knowledge commons and public participation. Using the Governing Knowledge Commons framework (Frischmann, Madison, & Strandburg, 2014), we examine international and national guidance documents that address AI in the context of development, including the outputs from the UN and the World Economic Forum, and the collection of documents under the Executive Order on Artificial Intelligence in the US. In this analysis we will be looking for the extent to which existing definitions focus on the social and the technical sides of AI and for how various attributes of AI, such as data, infrastructure, models, and communities around it are envisioned and conceptualized in these documents. We then compare these discussions to public participation literature on AI, exploring how these approaches align with the principles of equity, safety, responsibility, and localization.

We conclude by highlighting areas for empirical research that could strengthen responsible and equitable AI development and by presenting several key questions for further discussion on the barriers to collective action and public participation in AI.

Conservation treaties are agreed-upon institutional (rule) configurations designed to create mechanisms that share information on global biodiversity conditions and are adjustable to changing social and ecological conditions so that core treaty objectives are continuously met. Since adjustments in treaty rules can lead to fragilities elsewhere, the goal is to create a governance system that can cope with change while maintaining core conditions within acceptable parameters, e.g., to prevent species extinctions and biodiversity declines. Polycentric institutional design is thought to provide such flexibility. However, few studies have examined whether treaty rule configurations are indeed polycentric and how this may influence adaptability to change.

Polycentric systems are defined by three overarching attributes: (1) they consist of many centers of decision-making; (2) they are governed by a single system of rules that can be institutionally or culturally enforced; and (3) they foster contestation of ideas, methods, and “ways of life” that lead to the emergence of a spontaneous social order that fosters the ability to change. The Logical Structure of Polycentricity (LSP) framework provides indicators for each of the three attributes (Aligica & Tarko 2012). We couple the LSP framework with the Institutional Grammar 2.0 to measure these indicators and explore the polycentric elements of the formal institutional design of four international conservation treaties. We then provide a categorization of the degree of polycentricity evident in each treaty design with a particular focus on how the identified elements may privilege certain actors with more power to shape conservation governance, and how these power dynamics may affect trustworthiness and collective action within treaty regimes.

Citations: Aligica & Tarko (2012). Polycentricity: from Polanyi to Ostrom, and beyond. Governance, 25(2), 237-262.

Our paper advances two arguments. First, we argue that the existing law and data-political economy in the European Union perpetuates commodification of open data. Instead of enabling a remix culture for shared knowledge production, legal frameworks enable capture of open data by actors with more infrastructural power. This excludes individuals and communities (i.e the ‘real’ data generators) from deriving value out of open data.
We illustrate this argument through three ‘rules-in-use’ for the open data ecosystem. At the constitutional level of regulatory law, literature from critical data studies illuminates the gaps between the ‘imagined’ and the actual beneficiaries of open data initiatives. At the collective choice level where open data and content licenses serve as private legal ordering, license terms are misused during data re-use for training AI models. At the operational level, we discuss commonswashing by Big Tech platforms, for e.g., Google’s Data Commons project.
Second, we propose law and policy solutions that recognise the relational and ecosystemic nature of open data. We argue that to unlock the potential of polycentricity, we must recognize the multi-faceted role played by the state in an open data commons – as an enforcer/sanctioning authority, but also provider, consumer, endorser and curator of open data. We also argue that because data re-use is impacted by copyright, privacy and data colonialism, existing open data and content licenses need to include more community preferences. Finally, we argue that the enforcement of such community preferences requires new legal institutions, to ensure more equity in the distributional impacts of open data. These suggestions can shift the perspective from governance of open data as common property, to governance of the open data ecosystem as a commons that preserves democratic values.

Purtova, N., & van Maanen, G. (2023). Data as an economic good, data as a commons, and data governance. Law, Innovation and Technology, 16(1), 1–42; van Loenen, B., et al. (2021). Towards value-creating and sustainable open data ecosystems: A comparative case study and a research agenda. JeDEM - EJournal of EDemocracy and Open Government, 13(2), 1–27; Contreras. J. (2017). Leviathan in the Commons: Biomedical Data and the State. In Strandburg, K., Frischmann, B., & Madison, M (Eds.), Governing Medical Knowledge Commons (pp. 19-45). Cambridge University Press; Benhamou, Y., & Dulong de Rosnay, M. (2023) Open Data Commons Licenses (ODCL): Licensing Personal and Non Personal Data Supporting the Commons and Privacy.

Schlager & Ostrom (1992) look at tangible commons property as a bundle of rights. We look at how digital commons can also be understood as a bundle of rights. Free and open licenses have frequently been considered to form a bundle of rights which define roles, and segment who is included and who is excluded. Few have actually tried to analyze properly this assessment. Douglas (2011) showed that the key element for free and open source software is not ownership, but access and use, delineation of who has rights and duties. We argue that for digital commons, most of these rights are not based on ownership or copyright, and propose to extend the bundle of rights beyond open and free licenses.

Digital commons are a subset of the commons, where the resources are data, information, culture and knowledge which are created and/or maintained online. They are shared in ways that avoid their enclosure and allow everyone to access and build upon them (Dulong de Rosnay & F. Stalder, 2020). Free and open licenses define the resources' shared rights from a copyright perspective, but the resources’ governance rules result from a combination of licenses with other written and unwritten rules.

We draw from Douglas and include governance documents other than licenses, such charters and code of conducts, to discuss the pertinence of using the bundle-of-rights framework to analyze governance of digital commons beyond copyright.

We achieve this by investigating the rules-in-use for three projects - the operating system Debian, the food transparency database OpenFoodFacts and the communication protocol Matrix. The bundle of rights may include the rights to represent the collective, to take certain decisions, to manage conflict, to redirect the project, to change the license, to ensure care, etc.

References

Douglas, D. M. (2011). A bundle of software rights and duties. Ethics and Information Technology, 13(3), 185–197. https://doi.org/10.1007/s10676-010-9229-3

Dulong de Rosnay, M. (2016). Peer to party: Occupy the law. First Monday, 21(12). https://doi.org/10.5210/fm.v21i12.7117

Dulong de Rosnay, M. & Stalder, F. (2020). Digital commons. Internet Policy Review, 9(4). https://doi.org/10.14763/2020.4.1530

Guillier, V. (2024), Au-delà de la propriété ? Vers de nouveaux faisceaux de droits pour étudier la gouvernance des communs numériques, communication at the “TIC.IS
TIC, Information et Stratégies” conference.

Schlager, E & E. Ostrom, E. (1992), Property-Rights Regimes and Natural Resources: A Conceptual Analysis. Land Economics Vol. 68. n°3, p. 249-262.

Origin products result from the interaction between humans and the surrounding ecosystem. They are unique for a combination of factors, including a shared inter- and infra-generational knowledge resulting in specific producers’ choices to manage environmental localised resources. Collective action dynamics aliment the place-based reputation of origin products, the complex and nested shared resource. If exposed to unregulated access and use, this resource suffers from collective action problems potentially leading to its depletion. In Europe, the place-based reputation can be protected through the Geographical Indications (GIs) regime. Applicant producer groups must then engage in rulemaking, regulating the access and use of the resource, while setting sufficient restrictions against free-riding. When this rule system operationalises a commons type of management, market and non-market spillovers are generated for the benefit of the producers and of a wider local community. The coexistence in GI origin products of private and public interests legitimates the intervention of external actors (namely the State and third parties) for ensuring that rulemaking avoids arbitrary exclusions. However, external actors’ involvement is not harmonised as to the procedure, approaches and product classes, even within the same Member State. Therefore, the trustworthiness of the GI as an intellectual property tool characterised by well-defined policy objectives can be jeopardised. In this contribution, I will present the methodological and theoretical implications of using the GKC framework to analyse rulemaking in GI settings. Using a simplifying tool called the ‘Actors-Process-Outcomes’ (A-P-O) approach, I will first discuss the introduction of additional variables in the GKC framework, focusing specifically on those related to external actors. Then I will argue that this addition can be a valuable entry-door for discussing polycentricity from a legal-institutional perspective. Finally, I will show the practical implications of this approach through targeted case studies involving French and Italian GI origin products. [1]

[1] Guerrieri F., Governing governance: Collective action and rulemaking in EU agricultural and non-agricultural geographical indications (2023), PhD thesis, University of Amsterdam.

Low-lying coastal communities are disproportionately vulnerable to coastal climate hazards that jeopardize livelihoods, health and wellbeing, heritage and connection to place, ocean-reliant economies, and critical infrastructure. To support communities in advocating for and making informed decisions about climate resilience in places they live, work, play in and rely on, ensuring information and data are accessible and usable for diverse users is critical. But how is this actually achieved? This presentation will describe a multi-pronged approach being implemented as part of a trans-disciplinary, community-engaged coastal climate resilience project in New England working waterfront communities geared at democratizing access to and the usability of information. This approach includes 1) the development of an ‘accessibility instrument’ against which to evaluate information/knowledge/data; 2) the creation of multimedia communication products to help researchers translate coastal resilience tools and data being produced for this project to heterogeneous audiences; 3) semi-structured community focus groups on data accessibility and information/knowledge gaps. Though this project is still in its early phases of implementation, we will discuss barriers and drivers to information accessibility internal and external to researcher’s sphere of influence; challenges and successes related to the co-development of relationships, networks, information; and themes including but not limited to data ownership in frontline communities, the hyper-localization of information, and the influence of researcher positionality and community context. We welcome feedback, discussion, and stories from attendees related to their own experiences making information more accessible to diverse audiences.