Bias in Algorithms

Brian Farmer
2 min readMay 27, 2021

--

The same issue that has plagued every topic this term is that the tech just isn’t up to snuff yet. The algorithms that engineers create are great and they offer humanity shortcuts and time saving strategies. The existence of them has probably significantly increased human productivity. However we all know that sometimes shortcuts only save time because you’re missing a very valuable piece of the process.

Algorithms allow us to skip over many of the hundreds of variables that make up a person’s life and personality. They can reduce a person to only a few key items meant to organize them into specific groups. We use these groupings to decide their fate and this can only result in people suffering undue consequences due to the nature of shortcuts. Of course the nature of these shortcuts are not limited to algorithms but to humanity in general. It takes time to get to understand someone and their goals or passions. Many people like to believe they can predict the value of a person based on a first impression. I can guarantee that this is a bunch of nonsense. No human has learned to read minds and understand the complexity of the genetic and environmental influences on personality to an exhaustive degree. This human shortcut to understand things quickly is based in the evolution of survival. We had to make quick judgements in order to survive the wilderness. This mental shortcut stuff is still a problem with us and it translates into our technology. Human shortcuts upon human shortcuts in order to save time and make quick judgements. While the root of this behavior is survival, the modern outcome is usually bias, discrimination, and prejudice.

How do we fix this? It’s not simple and I don’t believe that it’s entirely possible. Humanity can’t be entirely rid of its brain mechanisms. While we can ignore them with considerable effort and education, not everyone is willing to put in the work. Some people suggest having a more diverse tech population. While having a diverse groups of engineers would be great, it would not fix bias because, regardless of their unique perspectives, they are still human. The solution would be akin to much more complex algorithms created by people who understand their own bias and know how to shut it down in the pursuit of data. This requires education in math, science, and human behavior that for most would be far too much.

--

--