Artificial intelligence can be understood as a toolbox of different technologies that are linked to far-reaching processes of social transformation. The term also functions as a projection surface for technological, social, and economic hopes – ranging from quasi-religious promises of salvation to the transhumanist utopia that humanity, through science and technology, could transcend its biological limits. In this vision of redemption, there is a dystopian moment in which a machine – most vividly embodied in the image of the Terminator – threatens to bring about the end of the world. We also learn from this that it is a mistake to anthropomorphize AI, that is, to approach it as if it were human. In reality, AI processes data, recognizes patterns, and is by no means a moral actor. What easily fades into the background and risks being forgotten: AI is the product of extensive human labor – often invisible, often exploited.
To do justice to this reality, Rainer Mühlhoff argues for a changed understanding of artificial intelligence. Instead of conceiving AI as an embodied entity or a mere vision of the future, its present and complex entanglement with human labor and social organization – beyond classical human-machine interactions – must be made visible. He therefore proposes to think of AI as “Human-Aided AI,” that is, as a networked human-machine system that neither exists nor functions without human involvement. (Mühlhoff, 2023, Die Macht der Daten)
Especially in the context of AI systems such as large language models, it becomes evident that technological processes are deeply embedded in social power relations. In their article Between Subjectivity and Imposition: Power Dynamics in Data Annotation for Computer Vision, Miceli, Schuessler, and Yang emphasize that the assignment of meaning to data is often portrayed as a purely technical task. Yet the systematic categorization of relevant information in datasets is not a neutral technical procedure, but an act of assigning meaning. In fact, it constitutes a powerful act with far-reaching effects on individuals and society. (Miceli, Schuessler & Yang, 2022, Between Subjectivity and Imposition) The Taylorist notion that intellectually demanding tasks in AI development can be separated from simple, fragmented work optimized through processes thus proves to be untenable.
The overarching concept of platform labor encompasses those services in the AI sector that are mediated or provided via web-based platforms. Various forms can be distinguished, such as click work, microwork, data work, or ghost work, although their boundaries are fluid and must be redefined depending on the context. These terms also point to precarious working conditions, the absence of rights, and the often invisible dimension of human labor that makes digital systems such as AI possible in the first place. Platform labor is, on a global scale, part of a neoliberal economy in which work is rendered insecure, de-bordered, and flexibilized.
The focus of our evening will therefore lie both on making visible and acknowledging the human component in “Human-Aided AIs,” as well as on the working and living realities of digital laborers.
Free Admission
More information here: https://scads.ai/event/pillars-and-umbrellas/the-invisible-faces-of-ai/
