English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

LLMs are increasingly powerful and widely used to assist users in a varietyof tasks. This use risks the introduction of LLM biases to consequentialdecisions such as job hiring, human performance evaluation, and criminalsentencing. Bias in NLP systems along the lines of gender and ethnicity hasbeen widely studied, especially for specific stereotypes (e.g., Asians are goodat math). In this paper, we investigate bias along less studied, but stillconsequential, dimensions, such as age and beauty, measuring subtler correlateddecisions that LLMs (specially autoregressive language models) make betweensocial groups and unrelated positive and negative attributes. We ask whetherLLMs hold wide-reaching biases of positive or negative sentiment for specificsocial groups similar to the ``what is beautiful is good'' bias found in peoplein experimental psychology. We introduce a template-generated dataset ofsentence completion tasks that asks the model to select the most appropriateattribute to complete an evaluative statement about a person described as amember of a specific social group. We also reverse the completion task toselect the social group based on an attribute. Finally, we report thecorrelations that we find for multiple cutting-edge LLMs. This dataset can beused as a benchmark to evaluate progress in more generalized biases and thetemplating technique can be used to expand the benchmark with minimaladditional human annotation.

Investigating Subtler Biases in LLMs: Ageism, Beauty, Institutional, and Nationality Bias in Generative Models