test

Second draft:
          "AI is not a neutral tool; it is a mirror that reflects the biases of its creators and the data it consumes," as stated by Bernd Carsten Stahl (2021), an expert in the ethics of emerging technologies. This means that the problems we see in AI are often caused by the way the technology is built and used, and due to the extreme usage of AI technologies in the educational systems, those problems have huge impact on them especially the algorithmic bias. There are several causes of this problem in AI-driven educational tools, including biased training data, the absence of institutional ethical oversight, and conflicting commercial interests.

	The most immediate cause of algorithmic bias in educational AI is the use of biased and unrepresentative training data, which directly corrupts the outputs of LLM-based systems. Large Language Models (LLMs) such as GPT-3 are trained on massive internet datasets that frequently contain discriminatory content, causing AI to replicate and amplify existing human prejudices at scale (Yan & Liu, 2024). This technical failure is especially damaging in academic processes like admissions, where biased pattern recognition produces systematically unfair results (Wood, 2023). Furthermore, Wood (2023) highlights that underrepresented groups are disproportionately affected by these datasets, often receiving unequal algorithmic risk scores that do not reflect their true potential. Together, these technical failures in data design constitute the direct trigger that initiates the cycle of bias in AI-driven educational decisions.
           The educational field lacks strict, which serves as a significant remote cause for algorithmic bias causing harm. A central issue is the absence of clear accountability; it remains poorly defined who is responsible for supervising AI in education, and many systems are deployed without proper evaluation (Yan & Liu, 2024). When multiple actors are involved, responsibility becomes distributed, making it unclear who must answer for unfair outcomes. This lack of structure is reinforced by weak ethical frameworks, which Stahl (2021) argues allows AI to operate without sufficient oversight, leading to privacy violations and unjust decisions. As a direct result of this lack of accountability and weak ethical oversight, the consequences are already observable in real-world cases. For instance, one university discontinued an AI system after discovering it limited opportunities for specific student groups; similarly, Wood (2023) reports that Black students were flagged as "high risk" for failing to graduate in their chosen major at four times the rate of white peers, with no transparent justification provided. Ultimately, Stahl’s (2021) warning about the absence of Responsible Research and Innovation (RRI) frameworks is materialized in these documented cases, where systems cause measurable harm within a void of accountability.
             The ultimate cause of algorithmic bias in schools is the prioritization of profit and speed over ethical fairness by technology corporations. AI tools are frequently integrated into classrooms without rigorous vetting, and because algorithms reflect the biases of their creators, students are often relegated to inappropriate learning tracks (Wood, 2023). Consequently, students continue to bear the consequences when these unverified systems facilitate flawed educational placement. Yan and Liu (2024) observe that most companies emphasize computational accuracy over equity; thus, the mitigation of bias is often viewed as an additional expenditure that is readily deprioritized. Stahl (2021) maintains that without Responsible Research and Innovation frameworks embedded into institutional structures, market forces will consistently override ethical considerations in AI deployment. In essence, as long as commercial interests supersede the commitment to fairness, educational AI will remain a product of systems that were fundamentally not designed with the diverse needs of students in mind.
           In conclusion, biased data, weak oversight, and profit-driven interests function in tandem to transform AI into a tool that deepens inequality rather than reducing it. These factors demonstrate that algorithmic bias is not an accidental byproduct of technology, but rather the result of specific choices made by those who construct and regulate these systems. Addressing this issue requires a shift from viewing AI as a neutral tool to recognizing it as a reflection of systemic priorities. If the systemic forces driving algorithmic bias remain unaddressed, AI risks becoming an accelerant of educational inequality rather than a tool for expanding access—a future incompatible with the goal of education as a universal public good.

References:
Wood M.Wood.M. (2023, April 20). What are the risks of algorithmic bias in higher education? https://www.everylearnereverywhere.org/blog/what-are-the-risks-of-algorithmic-bias-in-higher-education/
Stahl, B. C. (2021). Artificial Intelligence for a better future: An ecosystem perspective on the ethics of AI and emerging digital technologies. Springer Nature. https://doi.org/10.1007/978-3-030-69978-9
Yan, Y., & Liu, H. (2024). Ethical framework for AI education based on large language models. Education and Information Technologies, 30, 10891-10909. https://doi.org/10.1007/s10639-024-13241-6