The Eye of the Master: A Social History of Artificial Intelligence (M Pasquinelli, Verso 2023)
Much breathless commentary has in recent years been exhaled about the impact of ‘AI’ – the moniker hung on Large Language Models (LLMs) produced using artificial neural networks that mimic human speech and dialogue. These computational wonders are pitched – by Silicon Valley capitalists, enabled by an army of uncritical tech journalists and others – as transformational to all aspects of people’s lives. “Hype Or Reality: Will AI Really Take Over Your Job?” and “Why Robots Won’t Steal Your Job” are typical headlines from recent months, and online influencers like firstmovers.ai are urging us all to ‘be ready!’
Such a combination of hype, anxiety, and hucksterism fills me with skepticism. Up until recently, that skepticism was instinctual, not articulate. I could say I didn’t like these tools, nor could I see the value in them to the user, but I had trouble articulating exactly why so much of the discussion and reality of ‘AI tools’ makes me nauseous. Then I read Matteo Pasquinelli’s The Eye of the Master: A Social History of Artificial Intelligence and I know why.
Industrial age machines and manual labour
Pasquinelli charts the development of modern artificial neural networks as the most recent iteration of capital’s multigenerational effort to steal workers’ knowledge and crystallize that knowledge into machines; machines owned by, and thus providing all their productive benefits to, the capitalist class.
Early 19th century polymath Charles Babbage begins the journey by articulating a labour theory of the machine wherein he explicitly argues that machinery ‘imitates and replaces a previous division of labour’. Machines emerge (or one could say, are invented) ‘only after a coordination of tools has been tested and proved to be successful for production and cost reduction.’ Babbage’s Difference Engine – the first mechanical computer – was explicitly designed and made to automate so-called ‘hand labour’ of the time, that is, hand calculation of logarithmic tables essential to both 19th century logistics in trade, and later on, artillery targeting in war making.
The author rightly asks the obvious question about this automation of knowledge – who owns the rights to such a machine? Who benefits from the productivity gains of putting that machine to use? Questions Karl Marx answered in a subsequent reading of Babbage’s work in his landmark study Capital.
Information age and mental labour
Fast-forward a century to the post war ‘information age’ and computation had become more technically sophisticated and capital intensive, with the US Army being a principal funder. The rewards again syphon upwards to the owners of such machines, with IBM and other mainframe computer manufacturers becoming massive (and profitable) enterprises. Attempts at making ‘thinking machines’ were again focussed on automating labour, this time through statistical predictions as a stand-in for that slippery concept of intelligence. The author writes that, “AI has actually emerged from the automation of the psychometrics of labour and social behaviours rather than the quest to solve the ‘engima’ of intelligence.”
The information age, Pasquinelli shows, differs from the previous industrial age only in the fact that where the industrial age sought to convert human physical labour into machines owned by capital, the information age seeks to capture, measure, and transform human mental labour. Consider all the convenient digital technologies you use – graphic design suites, the word processor I wrote this review on, web browsers and search engines, and yes, AI digital assistants too. All of these tools replace some piece of human mental activity previously performed by a skilled worker: a designer, a librarian, a journalist, teacher, call centre worker, paralegal, etc. So-called AI tools continue that same trend, only this time, they attempt to capture vastly more ‘worker knowledge’ – culture, language, emotional response, patterns of social movement and interaction. The complexity, and consequent cost, of such ‘machines’ has risen accordingly, though the author concludes that the core problem remains the same: “… any technology, and institutional apparatus, including AI, is a crystallisation of a productive social process. Problems arise because such crystallisation ‘ossifies’ and reiterates past structures, hierarchies, and inequalities.”
If the hype around AI makes you want to throw up or punch something (or both!), rest assured you are not alone, and your feelings make all the sense in the world. Capitalism makes us all sick, whether it wears a fancy digital glove over its clenched fist or not.
Did you like this article? Help us produce more like it by donating $1, $2, or $5. Donate
