The Netflix documentary “Coded Bias” deals with a major observation of modern computing: while computers may be tremendously fast, they are still dumb, and must be programmed by humans, flaws and all, to become anything more than dumb. Another major sub observation the documentary makes is that the bias carried within Computer Science is particularly tricky since it can be shadowed under the guise of the computer, i.e. it is maybe not always obvious when some biases are contributable to human held biases or universal, mathematical physical limitations. Something not mentioned too much in the doc that I have also noticed in Computer Science is a particular culture that pretends that all programming goals are objective, in other words, the computer is a digital object but programmers often train, or trick, themselves in thinking that they ought to, therefore, think like a computer. This is a difficult problem to get mad at as it is clear and obvious why such a culture thrived, but it is a wrong idea and a problem. The difficulty in “thinking like a computer” when programming is that you are unable to realize your own biases as thinking like a computer implies no biases, only objectivity. Unanalysed biases are dangerous and lead to un-objective results, and create a toxic positive feedback loop. As suggested by the doc, a proper approach might instead be to recognize the computer as a tool to humans, not something to be studied as an alternative to human thinking, but one that carries and often magnifies the subjective ideas of humans, in this way, it is instead extremely important to carry proper ethics and understand the nature of biases when programming. The tool will not filter these out as it is an objective machine that is being operated by a subjective being.
