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

The recent surge in the research of diffusion models has accelerated theadoption of text-to-image models in various Artificial Intelligence GeneratedContent (AIGC) commercial products. While these exceptional AIGC products aregaining increasing recognition and sparking enthusiasm among consumers, thequestions regarding whether, when, and how these models might unintentionallyreinforce existing societal stereotypes remain largely unaddressed. Motivatedby recent advancements in language agents, here we introduce a novel agentarchitecture tailored for stereotype detection in text-to-image models. Thisversatile agent architecture is capable of accommodating free-form detectiontasks and can autonomously invoke various tools to facilitate the entireprocess, from generating corresponding instructions and images, to detectingstereotypes. We build the stereotype-relevant benchmark based on multipleopen-text datasets, and apply this architecture to commercial products andpopular open source text-to-image models. We find that these models oftendisplay serious stereotypes when it comes to certain prompts about personalcharacteristics, social cultural context and crime-related aspects. In summary,these empirical findings underscore the pervasive existence of stereotypesacross social dimensions, including gender, race, and religion, which not onlyvalidate the effectiveness of our proposed approach, but also emphasize thecritical necessity of addressing potential ethical risks in the burgeoningrealm of AIGC. As AIGC continues its rapid expansion trajectory, with newmodels and plugins emerging daily in staggering numbers, the challenge lies inthe timely detection and mitigation of potential biases within these models.

Language Agents for Detecting Implicit Stereotypes in Text-to-image Models at Scale