SDGBiasBench reveals intrinsic SDG biases in VLMs driven by priors rather than evidence, and CADE mitigates them with up to 25% accuracy gains and 12-point MAE reductions.
Title resolution pending
10 Pith papers cite this work, alongside 13 external citations. Polarity classification is still indexing.
10
Pith papers citing it
13
external citations · external index
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
unclear 1representative citing papers
SPAGBias reveals that LLMs form nuanced gender associations with specific urban micro-spaces that exceed real-world distributions and produce failures in planning and descriptive tasks.
Claude Opus 4.6 fabricates more answers on Global North AI contexts than Global South ones, creating an exploitable vulnerability in AI control monitors.
Frontier multimodal LLMs perceive cities through a culturally uneven baseline where European and North American framings are closest to the model's neutral default, while prompting fails to recover human diversity.
Behavioral audit finds emergent, city-dependent racial steering in LLM housing recommendations that changes with user identity and preference context.
MobFusion fuses mobility networks into foundation models via three designs and reports improved performance on income, density, and crime prediction tasks using data from three U.S. metropolitan areas.
Analysis estimates 18.7% of Common Crawl documents contain geospatial information like coordinates and addresses, with little difference by language.
Multimodal framework using LLMs and VLMs with CAMERA fusion and ASTRA re-ranking outperforms text-only baselines on Local Environmental Observer Network dataset for spatiotemporal semantic search.
Responsible GeoAI for disaster mapping requires governance across data, applications, and society rather than algorithm improvements alone.
Proposes a user-centered approach with five goals and actionable strategies for ethical generative AI use in research, emphasizing utility evaluation and transparency over abstract principles.
citing papers explorer
-
SDGBiasBench: Benchmarking and Mitigating Vision--Language Models' Biases in Sustainable Development Goals
SDGBiasBench reveals intrinsic SDG biases in VLMs driven by priors rather than evidence, and CADE mitigates them with up to 25% accuracy gains and 12-point MAE reductions.
-
SPAGBias: Uncovering and Tracing Structured Spatial Gender Bias in Large Language Models
SPAGBias reveals that LLMs form nuanced gender associations with specific urban micro-spaces that exceed real-world distributions and produce failures in planning and descriptive tasks.
-
Geographic Blind Spots in AI Control Monitors: A Cross-National Audit of Claude Opus 4.6
Claude Opus 4.6 fabricates more answers on Global North AI contexts than Global South ones, creating an exploitable vulnerability in AI control monitors.
-
Culturally uneven urban perception in large language models
Frontier multimodal LLMs perceive cities through a culturally uneven baseline where European and North American framings are closest to the model's neutral default, while prompting fails to recover human diversity.
-
The Geography of Algorithmic Judgment: LLM Intermediaries, Place Identity, and Racial Steering in Housing Search
Behavioral audit finds emergent, city-dependent racial steering in LLM housing recommendations that changes with user identity and preference context.
-
Enhancing the Socioeconomic Understanding of Foundation Models with Urban Mobility
MobFusion fuses mobility networks into foundation models via three designs and reports improved performance on income, density, and crime prediction tasks using data from three U.S. metropolitan areas.
-
Quantifying Geospatial in the Common Crawl Corpus
Analysis estimates 18.7% of Common Crawl documents contain geospatial information like coordinates and addresses, with little difference by language.
-
Multimodal and Multiscale Spatial-Temporal Semantic Search and Recommendation with AI Foundation Models
Multimodal framework using LLMs and VLMs with CAMERA fusion and ASTRA re-ranking outperforms text-only baselines on Local Environmental Observer Network dataset for spatiotemporal semantic search.
-
Unbox Responsible GeoAI: Navigating Climate Extreme and Disaster Mapping
Responsible GeoAI for disaster mapping requires governance across data, applications, and society rather than algorithm improvements alone.
-
Beyond principlism: Practical strategies for ethical AI use in research practices
Proposes a user-centered approach with five goals and actionable strategies for ethical generative AI use in research, emphasizing utility evaluation and transparency over abstract principles.