Global Digital Data Governance | Polycentric Perspectives | Carolina A
www.taylorfrancis.com
This book provides a nuanced exploration of contemporary digital data governance, highlighting the importance of cooperation across sectors and disciplines in

(Reproducing this from Fission Talk)

Below I will add in my condensed chapter notes. Each chapter is available as its own paper on the book website 1 if you want to just pick and choose. Many of these chapters were presented as papers at the Internet Governance Forum in 2022.

Here is the TOC for your reference:

  • Chad KohalykOPMEnglish
    1·
    8 months ago

    Ch13: Polycentric Theory Diffusion and AI Governance

    • this chapter explains the attributes of polycentricity shaping the global implementation of AI technology.
    • some key insights of polycentric gov:
      • The Bloomington School explains how polycentricity has three features: (1) multiple centres of semiautonomous decision-making; (2) the existence of a single system of rules (be they institutionally or culturally enforced); and (3) the existence of a spontaneous social order as the outcome of evolutionary competition between different ideas, methods, and ways of life
        • NOTE: #2 are the “APIs” of polycentricity.
      • In polycentric governance, there is no single decision centre with ultimate authority
    • While some leading corporations have made more explicit their rules and procedures about AI, and are advocating for self-regulation, smaller organisations and individuals affected by Al technologies are excluded from central debates.
    • Developing AI demands constant data acquisition. This practice is not always aligned with and is even often in opposition to some basic principles of data governance worldwide, such as data minimisation, standardisation of the quality of data, and transparency of data use.
    • a single mistake made during development may be repeated millions of times due to automation, while no good way to trace the error data points.
    • Governing the use of personal data by Al is more than a technical or economic question; it also involves ideological competition and the negotiation of cultural interests. The fragmentation (Biermann et al. 2009) of several ideologies, such as technocracy and central planning-based forms of socialism, may reinforce the existing digital divide between and within the Global North and Global South.
      • Research shows that algorithmic fairness as understood within Western norms is not easily translated to, for example, the cultural values of people from India
      • when we accept common ethical principles such as transparency and fairness, the gap between perceptions and enforceability may increase due to the process of implementing these policies in local contexts.
    • Generally speaking, law faces enormous difficulties in Al regulation… Evidence shows that it is not easy either to translate code into law to address new social justice issues or to execute law in code to improve regulatory efficiency.
    • using code-driven automatic regulatory approaches to execute current laws does not necessarily guarantee efficiency.
      • ==NOTE: We want to prioritize justice, not efficiency, right? Both?==
    • In December 2021, the Dutch Data Protection Authority (Autoriteit Persoonsgegevens) announced a fine of € 2.75 million against the Tax and Customs Administration… [who] focused their attention on people with “a non-Western appearance,” … preventing them from claiming the childcare benefits they deserved. … Such risks can cross national boundaries due to global market forces.
    • The analysis of Lessig’s four regulatory modalities, traditional to cyberspace governance debates with the main categorisations of the Bloomington School of polycentric governance and the more fluid and global perspectives of polycentric theorising (Koinova et al. 2021), all suggest that there are many ways in which regulatory approaches can help understand AI tensions by stressing the sources of power that shape these debates.