Abstract
When the prevailing thoughts from the sectors of governments, businesses, industries, and even the academia are cheering the affordance and effectiveness of the latest developing technologies of artificial intelligence (AI), Pieter Verdegem and a dozen other scholars started to think from a different perspective – a critical evaluation of AI's impact on humanities and the rising issues on digital power, algorithmic domination, and social egalitarianism. The edited book AI for Everyone? Critical Perspectives provides such an insightful examination of the hidden logic, agenda and mechanism behind the scene. Overall, this book on the one hand challenges the power and ideology of artificial intelligence and tries to respond to AI's rising risks on the other.
AI does provide new business and technological opportunities and brings many conveniences to users/customers. AI can create efficiencies, upgrade industrial standards, improve automatic driving capabilities, and assist some people in need in finding new solutions… When employed properly, AI could create a lot of benefits for mankind. However, as the book editor Verdegem points out: ‘We are being told stories about AI as the ultimate innovation, transforming the ways we live and work … At the same time, however, analysis is revealing that AI itself is one reason behind intensifying societal problems and harms’ (p. 2). These negative impacts include discrimination, growing inequalities, bringing technological unemployment, and may even lead to the end of humanity! Authors urge the need to engage in these problems and encounter the difficulties that AI has brought to the modern world.
This book is organized in three parts consisting of 15 chapters. Each part is introduced and reviewed as follows:
On AI – humans vs. machines
Is AI a machine or human? Since the conception of body-and-machine mixed ‘cyborg’ was born, human beings have become more and more interconnected with the techno-embedded material world. From machinery arms, electronics, and digital data to brain-in-memory chips, almost everything alien to human beings can now coexist with us. When AI can think like humans, think rationally, act like humans, and/or act rationally, AI and its ‘robots’ and ‘bots’ babies will become the competitors of human beings. Will humans’ existing social order and institutions such as occupational structures be threatened by these ‘things’ and ‘intelligence’? With the advancement of AI into human bodies and souls, that is, man–machine-hybridization, will human beings become ‘post-humans’ or even ‘non-humans’? How do we make friends with these new-born AIs and co-exist with them?
Wolfgang Hofkirchner argues that the tension between humans and machines can be eased by instituting a mechanism called ‘combinations’ (instead of ‘conflations’ or ‘disconnections’) and adopts the idea of ‘digital humanism’ to deal with it. ‘The way out is the establishment of a relation through affirming both the identity of and the difference between, the two sides – as done by combinations. Combinations provide the proper basis for a humanism that is up to the challenges of digitalization – Digital Humanism’ (pp. 45–46). Andreas Kaplan argues although algorithms have problems dealing with ethical and moral thinking, machines are better at utilitarian, repetitive tasks. ‘Regulation and guidance are definitely needed in order to avoid such bias, to establish a good foundation for machine–human collaboration’ (p. 26).
Jenna Ng formulates an alternative understanding of creative AI by examining ‘how the computational terms of AI may be rationalized in a framework intelligible to humans’ (p. 49) and thus urges for such a formation of intelligibility that can be applicable to algorithmic systems. Dan McQuillan, on the other hand, thinks about the existence of a ‘post-algorithmic humanity’ and writes on post-humanism and mutual aid between humans and machines. ‘We are seeking an alternative AI that avoids the dehumanization induced by automated segregation. Where AI is an engine of injustice it is because it intensifies the reductiveness, representationalism, and universalism that privileges an existing social hegemony’ (p. 75). Instead of a technocratic solution, McQuillan argues that a complete socialization of the relations needs to be involved in order to seek for a new materialist AI.
On discourses and myths about AI
The content and form manifested in AI-driven media involve languages, usages, terminologies, pictures, video clips, artworks, data, platforms, and so on. They together create a structure of discourses, metaphors, ideologies, meanings, and myths about artificial intelligence. How do we understand these discourses and be able to decipher the myths about these depictions of AI? This book intends to enquire into these issues in part II.
In terms of understanding the hidden meaning delivered by AI, Rainer Rehak provides a constructive critique of the terminology generated and used in the AI discourse. Rehak ‘scrutinizes the AI discourse regarding its language and specifically its metaphors and points out the problematic consequences and contributes to the debate by proposing more fitting terminology’ (p. 89). As there are increasingly more researchers probing into the need for literacies in deciphering the hidden discourses created by digitalization and artificial intelligence, a proper understanding of the formation of AI discourse and the computer-generated language structure becomes important.
In the chapter entitled ‘AI Ethics Needs Good Data’, Angela Daly et al. postulate that ‘we present a politically progressive approach to AI governance based on “good data” which seeks to empower communities and progress the priorities of marginalized and disenfranchised groups worldwide’ (p. 104). A ‘Good Data Approach’ is proposed with four pillars that can function as a guide for an ethical and politically progressive approach to AI development, governance, and implementation. These pillars are Community, rights, useability, and politics. Good Data Approach is thus ‘situated in an ethical perspective to progress society … Good Data also can take place and be relevant to all stages in the data collection process, from the beginning to the end’ (p. 115).
James Steinhoff addresses the notion of ‘AI for everyone’ via the concept of social reconfiguration. By discussing AI machine learning and aspects of its materiality, two dimensions of reconfiguration are introduced – utility and feasibility. Steinhoff argues that ‘existing considerations of the reconfiguration of AI have primarily focused on utility and have largely neglected questions of feasibility. Feasibility is considered primarily in relation to the materiality of AI, or its concrete aspects which “set constraints on and offer affordance for use”’ (p. 124).
Benedetta Brevini critically examines the myth-making phenomenon in Europe. The biggest problem lies in ‘the construction of a discourse that makes us perceive AI as the solution to the major problems in our society, including the inequalities brought about by capitalism’ (pp. 151–154). Brevini points out the three existing myths in the discourses on AI in Europe. First of all, AI is a solution for humanity and capitalism's biggest challenges. Secondly, creating urgency and ‘preparing’ society – AI is ineluctable. And finally, AI surpasses human intelligence.
On AI power and inequalities
In the final part of the book, a critical evaluation of AI focusing on power dominations and economic inequalities is done by several macro-structural examinations and case studies. Cultural studies of Baudrillard's idea of simulacra are introduced to echo Wiener's cybernetic notion developed in earlier technological inventions. The current foundation of AI – algorithmic logic – is regarded as deeply rooted in instrumental rationality in the context of digital capitalism. AI is also explicated in terms of the biopolitics of discrimination and domination in the case of facial recognition. Furthermore, ‘platformization of labor’ and ‘data justice unionism’ have been proposed to examine the transforming relationships among AI, workers, and social movement organizations.
By referring to Wiener's notion, Carrie O’Connell and Chad Van de Wiele pose that ‘the potential for the abuse of cybernetic systems [is triggered] by external forces … the machine's danger to society is not from the machine itself, but what man makes of it’ (p. 195). They remind us that the subjectivity and creator must remain in the hands of humans and cannot be controlled by the ‘sublime or other godlike manifestations’ such as AI and algorithms. Jernej A. Prodnik regards algorithms as part of a narrow form of artificial intelligence. The ‘algorithmic logic’ is formed in digital capitalism and is predominantly under the control of powerful capitalist corporations. Prodnik argues, ‘Major ensembles of algorithms today are developed and owned by some of the biggest corporations in the world’ (p. 207). The logic of instrumental rationalization ‘will lead to further social atomization, reification, domination, and alienation’ (p. 220).
Asvatha Babu and Saif Shahin, on the other hand, provide an analysis of algorithmic governance. They argue that ‘the prejudices they exhibit are a consequence of the systemic bias they are produced’ (p. 239). Social discrimination is produced ‘in terms of “historical biases” at first but eventually became a function of machine learning-related inaccuracies of the technology itself’ (p. 237). The algorithm, as the major form of artificial intelligence, forms the biopolitics of discrimination and domination long-established in the past.
Rafael Grohmann and Willian Fernandes Araújo explore how algorithmic management and surveillance lead to digital colonialism and therefore, function to create a global AI platform of Labor. They argue that ‘global AI platforms accelerate the platformization of labor from the process of “taskification of labor”’ (p. 251) and propose that ‘there is an AI colonialism reinforcing North–South inequalities from a platform labor perspective’ (p. 262). The final chapter in this book is presented by Lina Dencik urging for a labor perspective on AI governance. AI limits laborers’ ability to influence decisions that govern their lives. The solution is to promote the so-called ‘data justice unionism’. Dencik describes this concept as ‘a form of social justice unionism that engages with data-centric technologies as firmly situated within a workers’ rights agenda and that approaches AI governance as informed by the labor movement in solidarity with other social movements’ (p. 268). In facing the growing strength of AI, the authors of these two chapters all suggest the importance of labor power formation. They propose labor movements and cooperations between different groups in order ‘to engage with data justice in a meaningful way’ (p. 280).
Radical democratization of AI
AI has brought many positive dimensions to life, and at the same time, creates many worries about humans’ fundamental existence of subjectivity and humanism. Discussions on post-humanism, digital humanism, and post-algorithmic humanity in this book can reveal some concerns about this. With the emergence of AI, algorithms, and robots in our modern world, plenty of social issues and problems relevant to AI are raised. These include job market and employment, peace and democracy, power and domination, control and monitoring, bias and discrimination, as well as growing digital inequalities. The authors of this book know that humans will not be able to live without AI in the future. Human beings have to encounter it, co-exist with it, and achieve mutual benefits.
Pieter Verdegem, the editor of this book, proposes ‘If we want to establish AI that transforms society for the better and enables human emancipation, we need a radical democratization of AI’ (pp. 13–14). What does this mean? As Verdegem points out, this concept involves three major principles, as follows:
AI should be accessible to everyone: Nobody should be left out of the use of AI because of differences in race, gender, class, and other distinctions. Verdegem postulates: ‘In a decent society, all persons should have broadly equal access to the advantages and possibilities being created by digital technologies such as AI’ (P. 13). The digital divide in the AI era needs to be minimized.
AI should represent everyone: ‘In a decent society, all members should have a say about what type of AI is being developed and what services are being offered’ (P. 13). AI should not only serve the needs of the dominant class and high social-economic status. It should also address the needs of other social classes and status groups.
AI should be beneficial to everyone: Verdegem points out: ‘Developments in AI should contribute to the well-being of everyone in society. This matches with Wright's (2019) ideas of community and solidarity, which are crucial because of their connection to human flourishing and of their role in fostering equality and democracy’ (P. 14).
In sum, chapters of this book look into the development and influence of AI or algorithms on social, cultural, political, and economic consequences. Authors examine the changing realities, offer criticisms, and provide opinions on how to encounter AI. Some authors even think about how to live with it peacefully. These are all valuable discussions for understanding the changing relationships between human and machine – people and AI. Verdegem's ‘radical democratization’ principles incorporate great human values such as egalitarianism, digital democratization, and human emancipation. However, whether these principles can be achieved in most realities is still questionable. Besides, it will be a big challenge to confine AI to restricted usages as AI has already found a position watching us closely!