Processing underlying algorithmic principle —rebellion to algorithm-induced ethical fallacy
In Ramon Amaro’s article “as if”, the algorithm-induced ethical fallacy such racism and racialization are unveiled through a computer vision recognition project made by Buolamwini. For Buolamwini, a failure to recognise black faces as coherent human objects demonstrates “a lack of diversity in the [data] training set [that] leads to an inability to easily characterise faces that do not fit the normal face derived from the training set.” At the same time she reminds us that ”whoever codes the system embeds her views… limited views create limited systems.”
In substance paradigm, it is generally believed that causality can not empower the attributes to dynamic organisation and structures. However, in the domain of computer and data science, especially in artificial intelligence, the causality within algorithm indeed plays a significant role in the representational forms of the data, which fundamentally influences the organisation and structures of the expected algorithmic outcomes. For instance, nowadays algorithmic fallacy is mainly caused by a causal system where the data reflecting on physical reality directly influence and regulate the dynamic organisation within the system. We could also say that we are merely using algorithmic intelligence to validate and iterate our current social and ethical realities without any speculative interference. To be more specific, a physical reality that cultivates racism will never engender an equal organisation because the system per se is feeding on the representation of inequality.
Techno-science feminist Donna Haraway delves deeper into the cause of current socio-ethical degradation by bringing some new thought in our very nature of being optically dominated. In her article called “The persistence of vision”,she writes, “The human eyes have been used to signify a perverse capacity — honed to perfection in the history of science tied to militarism, capitalism, colonialism and male supremacy — to distance the knowing subjects from everybody and everything in the interests of unfettered power.” Additively, the racial-biased algorithms can definitely contribute to the history of science tied to racism. There are other reasons for this cause apart from that racism is historically embedded in western culture. The primary one should be the excessive demanding from computer vision and machine perception. It is generally acceptable that we need algorithms to drive machines that can do such things to unload or alleviate our iterative labour in daily life. To enable basic perceptional functionality inside machines, machine recognition becomes the overarching function to be developed and realised. However, the analytical and interpretive nature of western science constantly normalise machinery visual perception solely by the “cold and dry” logicalization of 2D imagery based on pixels. Even with depth-camera which could detect 3D space, it is still remained in the application of detectable colour and pixel organisation. From a philosophical perspective, the methodology to enable machine perception is constrained by the dogmatic approach to locate visible, graspable or perceptible elements, which strictly follows the doctrine of substance-based philosophy. An alternative approach could be derived from a completely different philosophical thought — process philosophy. Instead of focusing on representational elements such as pixel colours and propositional features, what if the machine is learning from the causal dynamic of those elements. Frankly speaking, what if the recognition and categorisation of machine perception are not based on specific features such as skin tone, silhouette and etc, but built on an aggregated learning results of how those elements are changed, transformed and directed, or it could also be concluded as the visionary dynamicity within the algorithmic causal relations. Recently, Japan researchers have erected a speculative methodology on using fractal patterns to teach an image-recognition system instead of using photos of real objects. Apparently, the rough idea still need some elaboration, which will be demonstrated in the fourth chapter in my dissertation “Envisioning and visualising speculative process metaphysics : two-dimensional twofold generative graphic “Monad” regulated by asymmetrical diverging syntaxes with Taoist semiotic approach”.
Writ large, a process thought-based self-realising or self-engendering algorithmic system might shed light upon an alternative approach for our socio-ethical solidarity.