Reporting Bias
Understanding Reporting Bias Reporting bias occurs when certain patterns or perspectives are disproportionately represented in datasets, resulting in skewed outputs from models trained on this data. Language models (LMs) like RoBERTa and GPT-2 are especially susceptible to this issue, as the biases embedded in their training data can influence the model’s responses, reinforcing these biases […]
User Interaction Bias in Machine Learning
Understanding User Interaction Bias In today’s world, where information is abundant and easily accessible, understanding how it shapes user experience is key to creating more equitable and accurate digital interactions. By recognising and addressing user interaction bias, we can enhance fairness and inclusivity across digital platforms, ensuring that users are exposed to a broader and […]
Sampling Bias in Machine Learning
Understanding Sampling Bias Sampling bias is like representation bias, which arises from the non-random sampling of subgroups. Because of sampling bias, the trends estimated for one population may not generalise to data collected from a new population Mehrabi et al. (2021). Example of Sampling Bias in Machine Learning Sample Bias occurs when a correlation […]
Representation Bias in Machine Learning
Understanding Representation Bias In recent years AI systems are often making headlines for their failures. A notable example includes Facebook’s ad algorithm, which excludes women from seeing specific jobs or Google Gorilla, an image recognition algorithm which failed to label black females appropriately (Suresh and Guttag, 2021). These failures frequently trace back to bias in […]
Learning Bias in Machine Learning
Understanding Learning Bias Learning bias in machine learning arises from the inherent assumptions, design decisions, and optimization goals embedded in model development. These biases can shape how a model interprets and learns patterns from data, often resulting in uneven performance across different groups or contexts. Central to this issue is the choice of objective function, […]
Historical Bias in Machine Learning
Understanding Historical Bias A lot of the time algorithms fail even after following systematic processes and best practices for sampling data and building models. This is because the model reflects past societal or systemic inequalities, such as gender, racial, economic and other biases. This bias is referred to as historical bias which stems from the […]
Deployment Bias in Machine Learning
Understanding Deployment Bias Machine learning (ML) is increasingly pivotal in decisions that directly impact individuals and communities. These algorithms learn patterns from existing data and generalise them to new, unseen scenarios. However, challenges often emerge during deployment, where models face real-world complexities that differ significantly from controlled development settings. For example, while tested on balanced […]
Cognitive Bias in Machine Learning
Cognitive Bias Imagine navigating the complex world of decisions we face every day—choosing what task to do, interpreting new information, or deciding who to trust. Our minds often rely on mental shortcuts, known as heuristics, to simplify these decisions. While these strategies are efficient and practical, they can sometimes lead us astray, resulting in cognitive […]
Social Bias in Machine Learning
Understanding Social Bias Machine learning models are developed using training data that often mirrors the society they originate from. Since this data is derived from human behaviour and experiences, social biases inherently influence it. These biases—rooted in cultural, societal, and historical inequalities—manifest in the training data and, subsequently, in the models, potentially perpetuating or amplifying […]
Measurement Bias in ML
Understanding Measurement Bias As artificial intelligence (AI) and machine learning (ML) increasingly integrate into society, concerns about the systematic inequalities these technologies may introduce or reinforce have grown significantly. Measurement bias, a key contributor to these disparities, arises when data collection or labeling processes fail to accurately capture the characteristics of different groups. According to […]