Apparent Irrationality in Technological Development – W3

With the advancement of AI technology, we often ask ourselves about the purpose and utility of it. Moves made by large tech companies can seem obscure or unrelated to the usage of AI amongst the working class. We can imagine a wide range of use cases where AI technology could have a legitimate use value, but the last few years imply a worrisome future where the major relevant companies involved in AI have a concerning level of control over media and information. As we approach a new horizon in AI technology, we must remember the past in order to understand the future.

Karen Hao’s Empire of AI speaks at length about the subtext lying beneath the development of AI systems going as far back as the creation of the name “artificial intelligence.” The most crucial points Hao illuminates is the larger economic and political contexts that characterized the history of this technology so far. We can start with a basic premise that those who control the means of production shape production to their interests. In other words, the essential purpose of a given technology is decided by the legal owner of that technology. Hao mentions the cotton gin and how its implementation deepened systems of oppression and power from the ruling class towards slaves in the Southern United States.

The major point that the development of the cotton gin exposes is that there exists a kind of tension or opposing force within the development of technology. On the one hand, we cannot deny that there exists a legitimate use value within the cotton gin which holds the potential for more efficient production that benefits people. On the other hand, the potential beneficiaries of said tech didn’t control its design or implementation. Thus, the purpose of the cotton gin, despite its inherent potential, was to further the aims and profits of slave owners. The technology itself holds the contradictory nature of its existence and use, and the general power dynamic of society is conserved in its material existence. This is the general premise of opposing interests that Karl Marx named “class antagonism.”

Kate Crawford’s Atlas of AI points out that AI has been and is currently being trained on data that lacks any contexts. The process of training AI on text, images of objects, or identification datasets removes any and all context from the actual substance of the data itself. People are turned into objects whose essence is evaluated by mere appearance and inferential association. Especially with facial recognition training, people are stripped of their humanity and become an amorphous pool nothing. Soren Kierkegaard dubbed called this effect levelling. The main idea is that concepts that group everyone into a category that has no intrinsic features abstracts the humanity and subjectivity from people. Kierkegaard’s main gripe was the concept of the “public” because the public was never any static thing. Because it was nothing, it could be anything. Crawford is able to reiterate this same process as it has materialized in the development of AI systems.

So, what am I getting at here? Well, the primary idea is that none of the development and implementation is novel or surprising in the macro-scale picture of things. We can look to the past in which slaves were stripped of their humanity and the subsequent technological advancements of the American South’s productive forces conserved and perpetuated this power dynamic. What I am saying is that class antagonism, objectification, and levelling have been keystone parts of capitalist production for its entire existence, so AI development will follow suit in becoming a means of furthering the subordination of the many and the capital accumulation of the few. The material systems responsible for this are conserved in the process of training AI. AI, when directed by a ruling class, will be tailored to the interests of the ruling class. While it may display a legitimate use value that could brighten humanities future, it is being developed in a material context of subordination and objectification. The training sets that are objectifying and levelling people only exist because the prior political and economic context did the same thing.

The people have been used just the same way that the images and words of people are being used: inanimate, meaningless, and devoid of humanity. AI is just one more way in which the material reality of power in the West is conserved within the advancements in the means of production.

Despite my haunting tone here, we shouldn’t forget that all technological advancement has consisted of normal people just like you. People with the same material position as you went on to build rockets, computers, and even machine learning systems. All these technological advancements inherently contain a potential for good. They, alongside you, hold the power to create a brighter future for us all. While our outlook may seem bleak, with every progression in humanity’s ability to create and produce comes a greater potential of change. Just like the contradictory nature of technology, the apparent decay in working class power brings us more technology to use for working class prosperity.