Networked Society and Governance: Algorithmic Default?

Abstract

Civil society is increasingly digitized and virtual in many parts of the globalized world of today. The networked society and the invisible “second economy” (Arthur, 2011) which powers the developed and developing countries generate debates about the degree to which the benefits outweigh the potential hazards. Artificial intelligence (AI) powered by its machine learning underpin much of the digital networked systems, and “free” services such as search engines, paid for by the “tailored advertising” we get when we view webpages. Most now recognize that the helpful “suggestions” on the web are simply adverts personally targeted at individuals who have searched for information on a topic or visited a webpage with sponsored material and cookies.

There have been cases of major political misuse of data such as the voter manipulation by the Cambridge Analytica company. We are not just referring to the hacking and “fake news” used by some governments to influence the affairs of another country. Some organizations have used AI to cynically target consumers’ weaknesses, for example, in financial management (Larsson, 2018).

Perhaps more significantly the network technology is often promoted as having potential for improving civil society through “failsafe” or default forms of regulation using the embedded Apps in domestic equipment and algorithms in much the same manner it is suggested that automatic self-driving vehicles help to improve road safety by cautious driving and sticking to speed limits and so on (Cockburn, Jahdi, & Wilson, 2015, pp. 6–7). However, algorithms and the associated machine technology have also been described as a “black box” technology where even those people running the algorithms cannot always fully understand or explain how decisions are reached in diverse systems used to evaluate many things from medical care to credit rating and finance (Danaher et al., 2017). There are issues of the budding “surveillance society” emerging from the proliferating “intelligent” apps enabling corporate “spying” on our everyday lives as some hackers have done by tapping into baby monitoring systems in homes. In addition to hacking there are large power asymmetries involved as between commercial data users and the lay public who are often the data suppliers as their personal data are harvested each time the web is used.

Therefore, it is hardly surprising that, according to the Pew Research Center report by Aaron Smith, released in November 2018, over half of Americans surveyed found it unacceptable to use algorithms to make decisions with real-world consequences for humans. In the age of connectedness and the emergent internet of things many people are not yet ready to cede more control of their currently offline lives to current online technology. This chapter reviews arguments for and against algorithmic governance.

Machine learning systems may be efficient to a high degree without being unbiased in impact across different segments of society. AI may also be fully effective in its operation without even being fully understood because the decision-making is so arcane. Importantly, though, even for those systems that have some human mediation or supervision, societal regulation is aimed at ensuring ends and means are aligned with human social, political and economic justice and thus socially effective as well as being technically efficient. Consequently, these systems have to require socio-emotional as well as cognitive safeguards. Although levels of implicit trust may vary demographically as between say millennials and baby boomers, high levels of trust, accountability and a culture of moral integrity must still form the bedrock for societal benefits.

Citation

Cockburn, T. (2022), "Networked Society and Governance: Algorithmic Default?", Oberoi, R., Halsall, J.P. and Snowden, M. (Ed.) Contestations in Global Civil Society, Emerald Publishing Limited, Leeds, pp. 63-78. https://doi.org/10.1108/978-1-80043-700-520221007