AI utopias and dystopias
For decades, two of the most iconic global cultural representations of artificial intelligence were the gleaming armies of humanoid robots in Terminator and the all-seeing red eye in HAL9000 — programs with extraordinary computing power, tasked with performing tasks of enormous importance and responsibility, ending up rebelling against their masters and trying to annihilate them. Meanwhile, AI is portrayed as the key to modern technological transformation, capable of freeing humanity from work and of saving human lives as well as the entire planet. The opposition between utopian and dystopian immaginaries has fuelled public debate about AI for a long time, but it is a somewhat naïve position in the eyes of those who develop and use these systems. Then again, AI has evolved, and continues to evolve, in the wake of the expectations, immaginaries, and visions that shape sociotechnical futures defining what is “conceivable” and guiding investing, government policies, and legislative frameworks. Scholars such as Sheila Jasanoff and Shoshana Zuboff have underlined the relevance and performability of these immaginaries, whose contextual and contingent nature derives from processes of mutual information involving situated knowledge, practices and visions of the future. The concept of “futures in the making” is useful here. Susan Halford uses it to identify the set of possible futures that are actively shaped by present decisions and ongoing processes. According to this perspective, the practices and immaginaries of the present are fundamental to understanding and contributing to changing the trajectories of possible futures.
Problematising impact
Adopting this conception of sociotechnical transformations also helps with another conceptual operation approached reluctantly by most technology experts: problematising using the term “impact” when analysing how technology and society are intertwined. While reflecting on the impacts of AI, or any other technological advance, is an opportunity for interesting contributions, it is limiting on an analytical level. On the one hand, it suggests that we should treat technology as a fact, a monolithic, external object. On the other hand, it implies reasoning according to a deterministic and linear vision of cause and effect, even where these relationships cannot be found at all or when they are not sufficient to describe the dynamics at work. So we end up talking about the impact of technologies on society as if they operated independently of it. Problematising impact does not mean denying the role of technologies in the future (and present) of society. Rather, it involves expanding this idea by studying the sociotechnical processes that cause certain technologies to emerge and be transformed according to particular configurations and practices. Problematising impact means examining the sociotechnical assemblage of the components that constitute AI systems, i.e. the complex of human and non-human actors whose interaction, and whose power relations, shape technology through bonds of mutual influence and co-construction. Only by acknowledging the importance of the sociotechnical assemblage of components on the continuous redefinition of the effects, successes, risks etc. of AI can we appreciate how naïve it is to dwell on the impact of technologies on society.
Copiloting the future
An important game is being played — that of defining fundamental roles and responsibilities in the creation, management, and use of AI systems, as well as how we think about them and the relationship we allow them to have with us. Private companies such as OpenAI, Inflection, and Meta contribute to these definitions while protecting their own interests and promoting extensive use of their systems. Microsoft has decided to integrate the Copilot virtual assistant into the Windows operating system by adding a specific button on the keyboard. It might seem like a minor change, but Microsoft has not changed its keyboard layout in almost 30 years. And so this is proof of the strategic weight that AI has for Microsoft. A new concept is proposed of IT services — assistants, consultants, informants, secretaries. On the other hand, there are now numerous institutions and research groups providing a comprehensive picture of AI applications and their risks to society while also proposing standards, regulations and good practices. The greatest difficulties are how these proposals can interoperate between the various fields of application, the level of detail they can reach, and how to communicate their content at different levels of abstraction in order to engage all audiences, whatever their level of expertise. It is not possible to ignore this transformation. But it is worth reflecting on how these relationships may change since they establish new future horizons and redistribute the contribution of each component in each one.
Whether with Copilot, Alexa, Siri, or numerous other AI systems, humanity is now in a position to “copilot” the future. That means creating spaces and practices of collaboration between human and non-human actors, with awareness of the defining power that they all have with regard to the future. It means actively choosing which trajectory to follow among the many being proposed, as well as contributing to the proposal itself. It also means rethinking, both analytically and ethically, the role of non-humans alongside humans. Today, it is not easy to regard the contribution of an AI system to a decision as being decisive, and it is certainly desirable to insist on the involvement of human actors in automated decisions (the concept of “human in the loop”, whereby automatic procedures remain under the scrutiny of humans). In order that this mutual adaptation of skills and responsibilities takes place without excessive consequences for society, it is necessary to collectively exercise the ability to think about the future and steer the present in the most inclusive, ethical and sustainable direction for all.
Bibliography
Halford, S., & Southerton, D. (2023). What future for the sociology of futures? Visions, concepts and methods. Sociology, 57(2), 263–278. https://doi.org/10.1177/00380385231157586
Jasanoff, S., & Kim, S. (2015). Dreamscapes of modernity. https://doi.org/10.7208/chicago/9780226276663.001.0001
Zuboff, S. (2018). The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power. PublicAffairs.
Image: Photo by Compare Fibre on Unsplash
