This essay was submitted to the Immerse Essay Competition 2025 and it received a Merit Award with a scholarship worth 1000 Euros.
Provided underneath is the full submission along with personalised feedback received from the Immerse Essay Panel of Judges.

Machine Learning Fairness
Discuss the concept of machine learning fairness and the measures needed to prevent AI bias.
‘Nothing is more precious and therefore more valuable than the ability to decide.’
-Napoleon Bonaparte
These days, the power of decision-making lies not only in the hands of humans but also in those of rational machines– maximisers of utility and minimisers of cost. As propellers of a just society, we must, then, raise the questions- To what extent are these machine-made decisions fair? Who is held accountable if an algorithm makes an unfair, discriminatory decision? To what degree can a machine positively or negatively impact the level of fair decision making in our current society?
These questions are answered by an interesting concept- Machine Learning Fairness, or in other words, the capacity of a machine to make decisions that disregard ethnicity, race, gender or other inherent characteristics.
It is a novice’s misconception that because machines themselves are incapable of the human propensity for emotional or personal inclination, they are perfectly fair. This would entail that machines can not only make the most effective and profitable decisions but also the fairest– with a complete indifference to social class, economic status, gender or race. At last, a world where automated job recruiter algorithms hire the people most capable irrespective of gender, and prediction algorithms enforce justice disregarding race. A world where everyone has equality of opportunity and minorities have a chance to truly reach their potential after years of marginalization– the utopian setting of every fairy tale but unfortunately, a currently unachievable dream.
A machine may be able to consider a significantly larger number of variables during decision making, leading to an uncanny accuracy in predicting the future– but it is no oracle. It is, at best, a mirror. One that not only magnifies humanity’s strengths but also its weaknesses. An algorithm is trained on unfathomable quantities of data– a magnanimous assortment of our personal and collective histories to help guide the decision-making process. How, then, can we expect the algorithm to be fair in its decisions, to not consider race or gender when it is something humans have always done in the past? While the machines themselves may possess no inherent bias, the data they learn from, teaches them to do so. Historic trends of fewer women in STEM caused Amazon’s automated recruitment system4 to employ more men. Years of hate speech on the internet percolated into Twitter’s chatbot, Tay5 teaching it to share anti-Semitic, transphobic and racial content, further invigorating the intricate web of hate in the digital realm.Our algorithms must perform better than us, not only in efficiency but also in fairness. The answer seems fairly simple– diminishing biases in data. Data must be collected from a variety of different populations to represent different ethnicities, genders and sexual orientations. At each step, a diverse set of end users must be considered. And what better way to ensure inclusivity in the data than to include them in the development of intelligent systems? There needs to be a greater representation at every stage of development– not only in data but also the engineers, who craft these algorithms. The road to representation may not be straightforward, but it is crucial to embedding fair decision making in our systems. Afterall, nothing is more precious than our ability to decide.
Bibliography
Bonaparte, Napoleon. Military Maxims of Napoleon. 1796. Project Gutenberg, https://www.gutenberg.org/files/50750/50750-h/50750-h.htm.
De Cremer, David. “What Does Building a Fair AI Really Entail?” Harvard Business Review, 3 Sept. 2020, hbr.org/2020/09/what-does-building-a-fair-ai-really-entail.
Kraft, Amy. “Microsoft Shuts Down AI Chatbot, Tay, After It Turned Into a Nazi.” CBS News, 25 Mar. 2016, www.cbsnews.com/news/microsoft-shuts-down-ai-chatbot-after-it-turned-into-racist-nazi.
Mehrabi, Ninareh, et al. “A Survey on Bias and Fairness in Machine Learning.” ACM Computing Surveys, vol. 54, no. 6, July 2021, pp. 1–35. https://doi.org/10.1145/3457607.
Russell, Stuart, and Peter Norvig. Artificial Intelligence: A Modern Approach, Global Edition. 2016.
Smith, Craig S. “Dealing With Bias in Artificial Intelligence.” The New York Times, 3 Jan. 2020, www.nytimes.com/2019/11/19/technology/artificial-intelligence-bias.html.
Sullivan, John, and John Sullivan. “Amazon Recruiting – a Case Study of a Giant Among Children.” Dr John Sullivan, 24 Jan. 2022, drjohnsullivan.com/articles/amazon-recruiting-case-study-part-1.
Essay Feedback
Communicating Ideas:
Your essay effectively explores the critical question of machine learning fairness while raising key ethical concerns surrounding decision-making algorithms. The introduction sets a strong tone, engaging the reader right away. To enhance your argument, consider providing more definitive examples or case studies that illustrate both the positive aspects of AI and the potential consequences of biases.
Word Choice & Structure:
Your vocabulary is sophisticated and well-suited for discussing complex concepts such as AI bias and fairness. Phrases like ‘maximisers of utility’ and ‘mirror of humanity’ add depth to your writing. To further improve, ensure that sentence structures maintain clarity while conveying intricate ideas; occasionally, simplicity can strengthen the impact of your arguments.
Critical Thinking & Analysis:
The essay demonstrates a solid understanding of the implications of machine learning and AI biases. Your references to real-world events, such as the Amazon recruitment case, effectively illustrate the issues at stake. To strengthen your critical analysis, consider exploring more counterarguments or alternative viewpoints regarding AI bias and fairness, which would enrich the discussion.
Using Evidence & Inference:
You reference reputable sources throughout your writing, which support your points on machine learning fairness. Your citations of specific incidents regarding AI biases enhance your credibility. Moving forward, integrating a wider range of studies or contrasting perspectives may provide a more nuanced understanding of the topic.
Star Sentence:
"It is, at best, a mirror. One that not only magnifies humanity’s strengths but also its weaknesses."
This sentence poignantly encapsulates the dual nature of machine learning algorithms, illustrating both their potential benefits and inherent risks.
Congratulations on the merit award!