DEV Community

Dr. Carlos Ruiz Viquez
Dr. Carlos Ruiz Viquez

Posted on

It's a fascinating yet unsettling phenomenon - AI algorithms

It's a fascinating yet unsettling phenomenon - AI algorithms can inherit bias from seemingly unbiased data. This occurs when these algorithms are trained on historical texts that contain subtle grammatical and linguistic patterns used by biased authors. The issue lies in the fact that language is not always objective, and even the most well-intentioned texts can reflect underlying biases.

For instance, consider a dataset of news articles from the past, where the authors used terms like "aspiring entrepreneur" to describe women, but "successful CEO" for men. If an AI algorithm is trained on this dataset, it may learn to associate these terms with specific characteristics or traits, perpetuating existing biases.

Furthermore, the use of linguistic patterns like active and passive voice can also introduce bias. A study found that articles written in the passive voice were more likely to feature male authors, while those in the active voice featured female authors. If an AI algorithm i...


This post was originally shared as an AI/ML insight. Follow me for more expert content on artificial intelligence and machine learning.

Top comments (0)