STEMming Inequities: How Does Bias Impact STEM Education and Artificial Intelligence?



Intro and Interest

Welcome to my page! Watch the video above to learn more about my topic, and continue reading down below.


Source: Medium

Reminders from the United States’ segregated past are evident through the disparities in STEM education for Black and Latinx students, compared to white and Asian American students, as reflected in the homogeneous technology sector, and revealed by the human biases in artificial intelligence (AI)

Over the past few years, society has started to grapple with how much artificial intelligence perpetuates and amplifies human bias— with harmful consequences. The phrase “data does not lie” has a clear juxtaposition with how data encodes and amplifies racism. Until very recently, a query on “Black girls” still returned pornography, and searching for traditional Black names on Google has a higher probability of returning arrest records for those names compared to searches for traditionally white names (Yin et. al).

Yet, is artificial intelligence alone to blame? Or has the United States’ heinous and immoral past that has left its stain on today’s society to blame? The underlying problem with bias affecting artificial intelligence algorithms is the homogenous group of engineers that create the system, and the few students of color who are receiving a STEM education. 

By creating more STEM pipelines, educational reform, and individual contributions, a diverse technology workforce would mitigate human biases in AI. To truly understand this educational inequity, it is imperative to understand its origins.

The Origins

“Photo: Ninth grade students are seen in a segregated classroom in Summerton, S.C., in 1954. (AP Photo/Rudolph Faircloth)” (PBS).

Despite the advancements away from slavery through Constitutional amendments and the Civil Rights Act of 1875, a series of Supreme Court decisions focusing on racial inequities nullified the work of Congress (Landman 17). Under Jim Crow laws, Black people were regarded as second-class citizens, separated from their white counterparts in public accommodations, transportation, prisons, the military, and most importantly, education (Landman 18). 

In 1896, the Supreme Court sanctioned the legal separation of the races in its ruling in Plessy v. Ferguson, which upheld the constitutionality of racial segregation under Louisiana’s “separate but equal” doctrine (Plessy v. Ferguson). Segregated schools served as the cornerstone of the Court’s argument. The ruling, upheld by sentiments of white supremacy, and white people’s belief in Black inferiority, provided grounds to enforce Jim Crow law across the country (Landman 17). 

The de jure segregation of southern schools and de facto segregation of northern schools persisted into the 1900s. Reminder’s of the South’s slavery-ridden past continued, with school schedules in rural areas being based around cotton growing season (Adedapo and Kaplan). As a result, students in rural areas only attended school only 2-3 months out of the year, and across all schools, Black children received a substandard education compared to their white peers. 

In the North, schools were primarily ruled by de facto segregation, creating a great contrast in education results in northern and southern schools (Brown v. Board of Education of Topeka, Khan Academy). In 1935, Dr. Otto Klineberg, a renowned social psychologist, found that on average, the longer children attended school in the North, the higher their IQ was tested to be (Lambert). The data further demonstrated that Black and white children had the same intellectual abilities, but the environment and access to education was the true deciding factor. Granted, the industrialized, urbanizing North was generally more prosperous than the South.

Who Fought Against it?

The St. Louis Board of Education was picketed by the NAACP after the Board issued a modified enrollment plan which did not go far enough in integrating the schools (The74Million).

In 1909, after numerous eruptions of anti-black violence, the National Association for the Advancement of Colored People, or the NAACP, was founded as the nation’s premier civil rights organization. Key members of the organization, including distinguished sociologist W. E. B. Du Bois, journalist Ida B. Wells, and NAACP chief counsel Thurgood Marshall led the charge in a series of court battles that culminated into key victories. 

A significant case for the advancement of the Latinx community was Mendez v. Westminster (1946), a class-action lawsuit involving more than 5,000 Mexican American students in Orange County, California. The case ended segregated schooling for Mexican American and white students (Valencia 389, 411), and served as a precursor to a Supreme Court case that shook the nation, Brown v. Board of Education of Topeka (1954). The landmark case was argued before the Supreme Court by Thurgood Marshall, who deemed that segregated schooling violated the Fourteenth Amendment and its promise of equal protection of the law. The Court agreed with Marshall’s findings, overturned Plessy v. Ferguson (1896), and concluded that “separate educational facilities are inherently unequal” (Brown v. Board of Education of Topeka, Khan Academy).

What Has the Problem Evolved Into Today?

Source: University of Toronto Magazine.

Black and Latinx Students Are Still At an Educational Disadvantage. Nearly 60 years after segregation, there is still a color line in education. Two-thirds of minority students attend schools where they are surrounded predominantly by fellow minority peers, most of them located in inner cities and funded well below those in neighboring suburban districts (Darling-Hammond). An analysis for school finance cases across different states discovered that on every tangible measure, including qualified teachers and course offerings, schools serving greater numbers of students of color had significantly fewer resources than schools serving mostly white students (Darling-Hammond).

School segregation is also linked to housing segregation, which is a significant source of the racial wealth gap in the U.S. (Natividad). As found in a Federal Reserve System study, white families have substantially more money than Black families. The report states that “Black families’ median and mean wealth is less than 15 percent that of White families, at $24,100 and $142,500, respectively” (Bhutta et al). The belief that schools are the great equalizer between different economic classes has not held true, considering nearly half of funding for public schools is paid for through local taxes, generating significant differences in funding between wealthy and impoverished communities (Biddle).

Students attending high-poverty schools have fewer opportunities and limited access to a rigorous STEM curriculum that includes courses, like calculus and physics, compared to students who attend wealthier schools (Sawchuk).

Stop and think: does your school offer STEM courses like calculus, physics, or computer science? How might that put you at an advantage over other students?

The Unbalanced Tech Industry

Source: Christina Animashaun/Vox

The inequalities in STEM education are evident when observing the demographics of the technology industry. In 2014, Silicon Valley began disclosing its workforce demographics, revealing companies’ personnel to have an overwhelmingly white, male, and Asian population (Harrison). A number of those companies made commitments to advance in diversity, but the numbers are still stark, especially amongst technical workers- the engineers, data scientists, and coders who serve at the heart of the company. Since 2014, at Google and Microsoft, the number of Black and LatinX technical employees rose by less than one percent. The number of Black technical workers at Apple remained at 6 percent, despite Black people making up 13 percent of the US population (Harrison). Due to the monochromatic group of technical workers across companies, the people who create products look and think the same. While companies treat diversity as a marketing pawn, a diverse workforce actually can—and does—bring about positive change.  

Totals may not equal 100% because of rounding and options to list more than one race (Harrison).

A is for AI, B is for Bias

Source: Simon Montag/The Atlantic
Watch this 2 minute video to understand how machine learning is intertwined with bias.

At a time when AI systems are being deployed across many industries, people must become acutely aware of the risks of bias in their designs, and work to fix them. AI can help identify problems and reduce the impact of human biases, but it can also absorb those of the people creating them. For example, a criminal justice algorithm used in Broward County, Florida, mislabeled African-American defendants as “high risk” twice as frequently as white defendants. Or during a study at MIT, researchers found that facial analysis technologies had higher error rates for minorities, particularly minority women, potentially due to a non-diverse data set being used while testing the technology (Manyika). Additionally, predictive policing algorithms have a long history of being weaponized against the Black community. As writer Dorothy Roberts stated, these algorithms are “about making certain racial groups seem as if they are predisposed to do bad things and therefore justify controlling them” (Heaven). The reality is that AI does not work until it works for everyone.

C is for Change

Source: Appuals

Local initiatives to close the racial achievement gap in STEM have yielded successful results, attesting to the possibility of diversifying the technology industry, therefore expanding the population that contributes to AI. 

In 2015, Intel pledged $5 million over five years to create a pilot project in two Oakland, CA public schools— Oakland Technical High School and McClymonds High School. The project aimed to produce hundreds of computer scientists and create a pathway for students in STEM (“Intel”). According to the schools’ data, “students enrolled in Computer Science classes grew from 196 in 2015 to 2800 in 2017. Students enrolled in AP Computer Science went from 24 in 2015 to 416 in 2017” (“House Science”). The students’ success demonstrates that, with the proper support and funding, school districts can improve educational outcomes and prepare students for the workforce. 

Other tech companies like Google have created their own programs to support Black and Latinx students. In Google’s “Code Next” program, “a free, computer science education program that meets Black and Latinx high school students in their own communities,” they work to develop the next generation of students through exploring the benefits of STEM education. Code Next has seen comparable results to Intel with 91.5% of their most recent graduates attending college/higher education programs, and 88.4% of those graduates will be majoring in a STEM field (“Code Next”). 

It is important to note that there are many STEM fields in addition to computer programming and engineering. Some of these academic disciplines include astronomy, biology, chemistry, earth sciences, health sciences, mathematics, physics, and more! Are you interested in any of these STEM fields?


Source: Harvard Business Review

Through large-scale initiatives, including corporate STEM pipelines and legislation to improve educational standards in public institutions, we can narrow the racial achievement gap, and biases in AI can be minimized by having a more diverse workforce. As demonstrated through Intel and Google’s Code Next programs’ success, if corporations were to invest a small percentage of their resources into shaping the minds of Black and Latinx students, they would create a substantial impact. 

STEM PIPELINES: The proposed STEM pipelines should commence as early as middle or high school, and continue to guide students through college. By starting the programs early on, students will learn to value STEM education and obtain the necessary skills to succeed in their future careers. Additionally, the company’s support of students while in college is crucial, considering that underrepresented minority students have a higher likelihood than other groups of switching to nonscience majors and are less likely to complete a degree in STEM fields (Tsui). This concatenation of patterns demonstrated by minority students is astonishing given that “in comparison to their White peers, underrepresented minority freshmen are just as likely, if not more likely to enroll in science and engineering studies” (Tsui). These students are capable of pursuing STEM majors, but they lack effective support systems, affirming that corporate STEM pipelines should continue through college to ensure the greatest success of the program. Furthermore, these companies would see a return on investment as their workforce’s size and diversity would expand. 

GOVERNMENT LEGISLATION: In conjunction with corporate STEM pipelines, there is a need for government legislation to improve public schools’ academic standards, specifically those that are underfunded. If promoted at the federal level, and implemented by state and local governments, a rigorous STEM curriculum and qualified teachers in public schools— no matter the location or tax bracket— would create equal opportunity for all students, and ameliorate the racial inequalities in STEM education. 

What Can We Do?

While the macro solutions would have a greater impact on closing the racial achievement gap and mitigating human biases in AI, individual initiatives would make a difference as well.

On a personal level, I can:

  • Engage in fact-based conversations around potential human biases.
  • Demand quality STEM education in my local school district.
  • Encourage technology companies to test AI on a wider range of people.
  • Educate others on the racial achievement gap and biases in artificial intelligence.
  • Submit my research for publication to share my findings and solutions with a widespread audience.

Call to Action!

Thank you for reading! Please take a moment to provide positive or constructive feedback in the comment section below, and please leave a like if you enjoyed reading. I appreciate your thoughts on any section of my work. For those interested in learning more, I have linked additional resources down below. Share this project with anyone who may be interested!

Here are some guiding questions for you to answer:

  • What is something that you learned?
  • What might you do as a result of reading this page? Or how would you implement the micro solutions listed above into your daily practices?
  • Do you think the solutions I proposed are tangible?
  • Do you have any other feedback or questions for me?

Works Cited

Click here to view my works cited.

Additional Resources

My Research

Here are some good reads to follow up on the topics mentioned through my page:

Interested in learning more about residential segregation, gerrymandering, and more? Read “The Color of Law” by Richard Rothstein- A powerful, disturbing history of residential segregation in America America

Are you a fan of data or psychology? Read “Datacylsm” by Christian Rudder- A look at human behavior and implicit bias through data

Do you want to work on your own biases? Read any of these books recommended by educators and activists.

Want to test your implicit bias? Take this Harvard quiz.

If you have not already, please engage in the comments by providing feedback or asking questions.



Student at Head-Royce School, Oakland, CA USA.


  1. This is such a well-crafted project! They way you present this already amazing topic in such a professionally designed and easy to follow manner is humbling. In terms of solutions, I think not only are they tangible, but they have high promise in a profit-driven economy such as ours. Tech giants just have to realize how much potential there is to granting the opportunity of high level stem education to everyone. While research and development is fantastic for humanity, sometimes it goes entirely wrong. It is terrifying and almost satirically foolish that some people would try to integrate artificial intelligence into legal procedures, especially given the historical injustices that such AI would use as data. Overall, a fantastic project.

    1. Hi Slava,

      Thank you for your compliments and in-depth feedback! I am glad that you enjoyed my project. I appreciate that you mentioned our profit-driven economy as that is a big incentive to companies’ work, and tech pipelines can serve as a win-win for everyone. Additionally, when looking at the integration of AI into the criminal justice system, I understand why developers would see it as an unbiased solution, but given the obvious data, it is incredibly prejudiced. Thank you again for engaging!


  2. This project is amazing! It is very well written and organized. Diversity in artificial intelligence is an important topic. However, it is not often talked about so I’m glad that you chose to focus on it. The proposed solutions are definitely tangible and I will do my part by following your personal level solutions. Well done!

    1. Hi Blair,

      Thank you very much for your compliments, and I am glad that you enjoyed reading my project. I certainly enjoyed reading your work as well. It is great to hear that you will follow the personal solutions. Thank you for engaging!


  3. Your project is so cool! I loved how you laid out the history, current situation, and possible solutions in a digestible manner. I learned about this a little bit in my computer science class, and I think you synthesized it really well. I really appreciate your willingness to dig in and explain this in plain terms — a lot of people think that AI and computers must be unbiased because they’re not human, but we instill them with the same biases that we ourselves have when we train them or program them incorrectly.

    1. Sophie,

      Thank you for taking the time to read my work and for the compliments. I appreciate that your computer science class takes the time to discuss the biases in artificial intelligence, and you made a very astute observation— computers are definitely not exempt from bias. Hopefully, if you continue along the computer science track, you can work to shift these biases and improve AI. Thank you for engaging!


Leave a Reply