Artificial Intelligence Unconscious Bias

Can Artificial Intelligence Algorithms have unconscious biases like humans ? Well before explore that, let's understand what is the definition for unconscious bias.

Unconscious biases?are social stereotypes about certain groups of people that individuals form outside their own conscious awareness. Everyone holds unconscious beliefs about various social and identity groups, and these biases stem from one's tendency to organize social worlds by categorizing. It is not limited to ethnicity and race. Though racial bias and discrimination are well documented, biases may exist toward any social group. One's age, gender, gender identity physical abilities, religion, sexual orientation, weight, and many other characteristics are subject to bias (Navarro).

According to a paper conducted back in 2018 by MIT and Stanford University, researchers analyzed 3 commercial facial-analysis programs and they found biases related to skin-type and gender. They experimented these programs and found that for light-skinned men the identification precision was around 99.2%, while for darker-skinned women was between 76% to 80% (depending on the algorithm). The finding raised questions around the way the neural networks were trained, since in general, they learn based on huge datasets those were formed by 77% of males and 83% of white, so the Artificial Intelligence [AI] algorithms just extrapolated what they learned. (Hardesty, 2018)

According to Joy Buolamwini, Phd, a researcher in the MIT Media Lab's Civic Media group, first author of the paper, "What's really important is the method and how that method applies to other applications. The same data-centric techniques that can be used to try to determine somebody's gender, are also used to identify a person when you are looking for a criminal suspect or to unlock your phone, and it's not just about computer vision. I am hopeful that this will spur more work into looking at other disparities.". As disparities we can consider when trying to get a loan for a new house or a car, when getting credit from the bank, of those social media algorithms that offer "weird" and "unsolicited" things for you.

Like human unconscious bias, the lack of information or diversity on these neural network training, lead to the Artificial Intelligence Bias. Neural Networks is already embedded in various devices and services we consume in a daily basis, and when building them it is important to consider a more emphatic approach and diversity, if we want to ensure a fair world, going forward.

If you want to know more about this, you can find in the references below, or watch a Netflix documentary called: "CODED BIAS".

Thanks for reading.

Luis A. Urso

References:

Navarro, J.R. About Unconscious Bias. UCSF Office of Diversity and Outreach. Available at: https://diversity.ucsf.edu/programs-resources/training/unconscious-bias-training. Accessed on: Nov, 4th 2022.

Nardesty, L. 2018. Study finds gender and skin-type bias in commercial artificial-intelligence systems. MIT News. Available at: https://news.mit.edu/2018/study-finds-gender-skin-type-bias-artificial-intelligence-systems-0212. Accessed on: Nov, 4th 2022.