The survey paper focuses on the concern of bias in machine learning algorithms and what role they play in affecting the fairness and accuracy of AI systems. Biases can present themselves in multiple forms, including but not limited to sampling bias, historical bias and annotation bias. Such prejudice can lead to unfair practices in sectors such as hiring, criminal justice, and health care where they can be used to serve and reconsolidate existing inequalities within the society. Additionally, the paper discusses the various approaches that can be taken to reduce bias in machine learning including adversarial debiasing, regularization methods, and post-processing strategies like equalized odds. All techniques focus on reducing the impact of bias; therefore, helping make AI models more valid and dependable. Beyond the technical details, the paper underlines the ethical side of the issue, saying that “bias reduction in AI is less of a great technological advancement and more of a moral responsibility”. Indeed, ensuring fairness in AI is critical for developing trust and making sure there's an impression that all are treated equally. Given the scope of both identifications of biases and introduction of mitigation methods, this research underscores the development of transparent and ethical AI systems which serve the public without reinforcing some of the prevailing injustices.