King – man + woman = queen

This equation, emblematic of early Natural Language Processing (NLP), served as a hallmark in demonstrating vector representations of words through an algorithm called Word2Vec. In these models, the words-vectors capture information about the meaning of the word based on the surrounding words-vectors.

Groundbreaking in 2013, the technique behind Word2Vec is still fundamental to AI, although it has been rendered obsolete by technological advancements – notably by ChatGPT.

Poking fun at the binary essence of the ‘King – man + woman’ equation, I conducted an experiment. By training Word2Vec on a dataset comprising texts authored by prominent figures in gender studies, I aimed to expand its semantic understanding. With this tailored dataset, I tasked the model to uncover words closely related to ‘queen.’