Mapping the Unseen: Researchers Engineer the Body’s GPS System in the Laboratory

Scientists have generated human stem cell models which contain notochord -- a tissue in the developing embryo that acts like a navigation system, directing cells where to build the spine and nervous system (the trunk). Scientists at the Francis Crick Institute have generated human stem cell models1 which, for the first time, contain notochord --
HomeEnvironmentNew Study Reveals Culture and Historical Biases Embedded in Language

New Study Reveals Culture and Historical Biases Embedded in Language

 

A recent study published in Social Psychological and Personality Science sheds light on the deep connection between people’s attitudes and language within various cultures and across different time periods.

Researchers examined the relationship between people’s attitudes and language across 55 diverse topics such as wealth vs. poverty, dogs vs. cats, and love vs. money. They analyzed four text sources, including contemporary English writing, English literature from the past two centuries, and texts from 53 non-English languages. To gauge attitudes, they collected data from over 100,000 Americans using both direct self-reports and indirect measures based on reaction times, known as implicitly-measured attitudes.

The study found that significant associations detected by advanced AI language models like ChatGPT align more closely with implicit attitudes measured through reaction times rather than explicit statements of attitudes.

Lead author Dr. Tessa Charlesworth, from Northwestern University’s Kellogg School of Management, highlights the importance of understanding the societal representations embedded in AI technologies. She suggests that rather than only focusing on identifying explicit biases, efforts should delve into the underlying patterns in the training data and offer alternative associations.

Dr. Charlesworth also emphasizes that implicitly-measured attitudes are intricately intertwined with language, a primary medium for transmitting cultural values. Addressing and diminishing implicit biases effectively may require interventions that adopt a broader cultural perspective.

While acknowledging the correlational nature of the study, researchers intend to further investigate sociocultural influences. Dr. Charlesworth points out the varying correlations observed in different non-English languages and emphasizes the need to explore social and cultural factors that might explain the link between bias and language transmission.

This study lays a foundation for comprehending how attitudes subtly intertwine with language and communication systems over time, reflecting not only the present but also echoing through centuries past.