When so called scientists are indoctrinated by the ideologies of the day! The title and abstract of the research paper sound harmless! However, the content of the paper and its intention appears questionable (I only glanced at the paper)!
When dogmatic censorship is called science! Of course, cause celebre Timmit Gebru is cited (I blogged here, here about her)!
"... What’s needed is a shift away from reliance on the large, poorly curated datasets used to train machine learning models. Instead, the study recommends a culture that cares for the people who are represented in datasets and respects their privacy and property rights. ..."
"... 3.1 Representational concerns
In recent years there has been growing concern regarding the degree and manner of representation of different sociodemographic groups within prominent machine learning datasets. For example, a glaring under-representation of darker skinned subjects has been identified within prominent facial analysis datasets [Buolamwini and Gebru, 2018] and the images in object recognition datasets have been overwhelmingly sourced from Western countries [DeVries et al., 2019]. Zhao et al. [2018] found a
stark underrepresentation of female pronouns in the commonly used OntoNotes dataset for English coreference resolution. Stereotype-aligned correlations have also been identified in both computer vision and natural language processing datasets. For example, word co-occurrences in natural language processing datasets frequently reflect social biases and stereotypes relating to race, gender, (dis)ability, and more ..."
Here is the link to the referenced paper (written by authors affiliated by e.g. Mozilla and Google):
No comments:
Post a Comment