From Özgür Nari

On the class structure of AI and the reasons that a productive force appear as a destructive force of capitalism.

Machine Learning Algorithms running on Artificial Neural Networks and particularly reinforcement learning were real breakthroughs of 2010s, Generative AI and transformers are of 2020s, all of them developed and spread to various branches.

And nowadays they even talk about an “existential risk” created by AI as if capitalism itself was not the real existential risk for centuries. So we should talk about capitalist production process of AI and the class structure of it.

Hence, I will first try to analyse the path from machine to machine-learning via Marx’s analysis of machine while considering the alienation as the dual and complementary part of it. Datafication of society is a consequence of this process. The data extractivism of capital transforms the datafication into profit seeking capitalist production as well as parameterisation of the society. So the data gathered in and for the sake of the capitalist production is the input of the capitalist cycle and at the same time actively intervening and manipulating both the labour process data and consumer behaviour data.

The data was and is not “as given” in a capitalist production. It would be better to discuss this constant manipulation before “bias” and “ethical issues” in data gathering. AI and constantly produced and reproduced data sets transformed this problem into a new sphere. When we consider the machine learning algorithms using large dataset gathered through all cycles of capitalist social reproduction, it would be better to consider the alienation of total social labour and this new level of “manipulation”. I will try to discuss this problem and suggest “parameterization of society” vis a vis “datafication of society” considering the distinction between the social form of data collection that is capitalist data extractivism and the objective content of data gathering which is in fact a possibility to organize the social production in a different classless society. The key factor is the subjective side, that is the intersection of the producers of the AI architecture, algorithms and the data.

The evolution from machinery to AI is actually a social process that is structured by class struggle. So I will try to trace the historical development of producing “learning machines” as a class struggle and point out both sides of the development, i.e. capitalist side of AI research, from cybernetics to AI and the Soviet side of it. And finally, modern class struggle, workers involved in the production of AI versus Tech-giants…

Last but not the least, considering the rather futuristic scenarios of AI, I will discuss the limits of AGI and creating AI with ‘the ability to transfer learning from one domain to other domains’ namely “AI with the capacity to engage and behave intelligently in a wide variety of contexts”.

Özgür Narin is currently a union member and Assistant Professor of Economics at Ordu University. He graduated from the Electrical & Electronics Engineering Department of Middle East Technical University. He studied the capitalist production of science and technology, particularly innovation and the changing scientific labour process. His current research is on Artificial Intelligence, “General Intellect” and the alternative reorganization of social production and society. His writing appeared on Monthly Review and Science & Society