A Research Group From CMU, AI2 and University of Washington Introduces NLPositionality: An AI Framework for Characterizing Design Biases and Quantifying the Positionality of NLP Datasets and Models

Researchers’ positionality—their perspectives formed by their own experience, identity, culture, and background—influences their design decisions while developing NLP datasets and models.  Latent design choices and the researcher’s positionality are two sources of design bias in producing datasets and models. This leads to discrepancies in how well datasets and models function for different populations. However, by […]

The post A Research Group From CMU, AI2 and University of Washington Introduces NLPositionality: An AI Framework for Characterizing Design Biases and Quantifying the …

ai2 ai framework ai shorts applications artificial intelligence biases cmu culture datasets decisions design editors pick experience framework identity language model large language model machine learning nlp perspectives research researcher researchers staff tech news technology university university of washington washington