AIExplainer

What is historical bias?

A systematic error in data or models due to past prejudices or discriminatory practices

Historical bias occurs when data or models used in AI systems reflect and perpetuate past injustices or biases, often unintentionally, leading to unfair outcomes or decisions

Historical bias is like a distorted lens that affects how we view and interpret the world, much like how a biased historian might write a skewed account of historical events, influencing how future generations understand the past

A facial recognition system that is more accurate for white faces than for faces of people of color due to a historical bias in the data used to train the system

Historical bias can be used to describe how AI systems may discriminate against certain groups of people, such as racial or ethnic minorities, due to biased data or algorithms that reflect past prejudices

Some people may think that historical bias is intentional or that it can be easily eliminated, but it can be deeply ingrained in data and models, requiring careful examination and correction

The concept of historical bias has been recognized in various fields, including sociology, psychology, and computer science, as AI systems have become more widespread and their potential impact on society has become more apparent

systemic bias legacy bias embedded bias

Three products for different needs — explore what’s relevant to you.