Why Psychology and Bias are Important to Consider in Technology

By Rachel Ware

Psychology is an important part of technology development that fascinates me. Consider that even a simple button on a website has many components: the size, color, font, whether it changes when clicked, where it is placed on the page, and of course the consideration of whether it should be there at all. Every decision when creating technology is motivated by a reason, whether carefully planned or the whim of the creator. Pieces of technology have some features that are determined by the framework they are created in and some that are decided for the particular application. In this way, even the platform used to build technology has an influence on its form and the considerations of its creators.

This intersection of psychology and technology is a field with many names: Informatics, Human Computer Interaction (HCI), User Experience (UX) design, and Human Centered Technology to name a few. Psychology is a consideration not just in the design, but also in the usage of technology. You might have heard about this from the recent Facebook whistleblower that sites such as Facebook are actively measuring the psychological impact of their platform. Many companies test user experience regularly and update their platform accordingly, whether to make it easier to order a pizza or add completely new features. This is not good or bad in itself, but can be seen as negative if the changes involved are not seen to benefit the users.

Let's jump back to the example of a website. When the design of the website is not carefully designed and tested, it can be greatly affected by the biases of the creator. Different people will have different mental understandings of what a website or app should do and what is important to consider. The impacts of this can range from a not so great visual experience to seriously changing people's lives. Have you ever tried to look at a website on a phone and it will not display properly? Or maybe you have wanted to create an avatar to represent yourself but cannot find a similar hairstyle or your skin tone. Or consider that some ways of using components on a webpage can be unreadable by a screen reader used by someone who is blind or visually impaired. These are all areas that might be overlooked in the design of a website but will be noticed quickly by certain people because it applies to them directly. If someone doesn’t have a computer, has darker skin, or is blind, these problems can steer them away from a website or other technology where their experience was not considered. However, sometimes people do not have a choice. When technology is used without people opting in, it can have the same sorts of issues with bias.

You may have seen some news about the perils of using artificial intelligence (AI) in policing, where people have been wrongfully accused of crimes or it has been used to guess recidivism rates, and usage in predictive policing. If you have not heard about it, this may sound surprising. Doesn't AI learn on its own? How can it be biased? Well, AI is made by people, so bias can be introduced. One major place where this bias enters AI is in the training data fed into it. For the policing examples, there is already human bias in policing: from where police patrol, to what crimes are charged, to differences in sentencing; all of which further marginalize groups. By taking this biased data from past police practices and plugging it into an AI, the AI learns the biases and its results will be biased as well.

Another problem can be a lack of data on certain groups of people. Datasets used for AI which recognize faces have been known to be flawed with underrepresentation of women and people of color in the training data. This has led to these models being much less accurate at recognizing the differences between faces of people who are non-white and not male, and even failing to recognize darker skin tones altogether. These issues impact government and industry use of these technologies, whose impact on people’s lives cannot be taken lightly. There is also the issue of miscategorization of gender, especially for transgender or non-binary people, and using these technologies without someone’s knowledge could be harmful to them.

Treating technology as though it is not impacted by the people making it would ignore a large part of why it exists in the form that it does. It can also keep people from understanding why diversifying tech is so important, because the perspectives of the creators of technology are reflected in it. When we think about who is creating the technology we use and how their own biases impact what exists, it shows us how important it is to consider both the positioning of the creator and how our own experiences with the technology are being shaped.