analiza aiqbench
This research investigates how persona prompting influences the interpretive framing of multimodal large language models when analyzing urban visual data. The study reveals that while descriptive captions remain consistent across different user profiles, justifications and semantic tags vary significantly based on attributes like economic status and political orientation. These findings highlight a critical divergence between objective grounding and subjective reasoning in current AI systems, offering insights into bias propagation within multimodal perception tasks.
Persona Prompting in Multimodal Urban Perception: Descriptive Convergence and Interpretive Variation
arXiv:2605.29064v3 Announce Type: replace Abstract: This study examines how persona prompting shapes language generated by two multimodal large language models in urban perception, a setting for examining subjective interpretations of shared visual evidence. We organize outputs into three functional layers: descriptive grounding (captions), intermediate semantic layer (perception tags), and interpretive framing (justifications). Using approximately 60,000 persona-conditioned annotations from each of two MLLMs, Qwen3-VL and Gemma4, we find that captions converge strongly across persona profiles and show only small attribute-associated differences. Justifications vary substantially more: economic status produces the largest difference in both models, with political orientation and personality also prominent. Paired image-level comparisons confirm larger justification than caption differences for these three attributes. For perception tags, personas sharing the same attribute level produce more similar tag sets than personas with different attribute levels, with the largest separation observed for economic status. Exploratory topic analysis further suggests persona-specific evaluative emphasis. Across models, profile-pair similarity patterns are strongly correlated for all three output types, although agreement is lowest for justifications. Overall, persona prompting affects interpretive framing more strongly than descriptive grounding.
Abstract:This study examines how persona prompting shapes language generated by two multimodal large language models in urban perception, a setting for examining subjective interpretations of shared visual evidence. We organize outputs into three functional layers: descriptive grounding (captions), intermediate semantic layer (perception tags), and interpretive framing (justifications). Using approximately 60,000 persona-conditioned annotations from each of two MLLMs, Qwen3-VL and Gemma4, we find that captions converge strongly across persona profiles and show only small attribute-associated differences. Justifications vary substantially more: economic status produces the largest difference in both models, with political orientation and personality also prominent. Paired image-level comparisons confirm larger justification than caption differences for these three attributes. For perception tags, personas sharing the same attribute level produce more similar tag sets than personas with different attribute levels, with the largest separation observed for economic status. Exploratory topic analysis further suggests persona-specific evaluative emphasis. Across models, profile-pair similarity patterns are strongly correlated for all three output types, although agreement is lowest for justifications. Overall, persona prompting affects interpretive framing more strongly than descriptive grounding.
Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Human-Computer Interaction (cs.HC); Multiagent Systems (cs.MA)
arXiv:2605.29064 [cs.CL]
(or arXiv:2605.29064v3 [cs.CL] for this version)
From: Thiago H. Silva [view email] [v1] Wed, 27 May 2026 20:11:42 UTC (229 KB) [v2] Fri, 7 Aug 2026 19:02:33 UTC (305 KB) [v3] Thu, 24 Sep 2026 03:17:17 UTC (312 KB)
Text integral sub licența deschisă a publicației · 231 cuvinte
Subiecte conexe
Multimodal Large Language Models Predict Urban Safety Perception but Encode Non-Neutral Demographic Priors
arXiv · 1 sept. 2026
NVIDIA GTC Taipei at COMPUTEX: Live Updates on What’s Next in AI
NVIDIA · 21 mai 2026
Agentic vision: Building visual intelligence with Amazon Bedrock and MCP servers
AWS · 15 iul. 2026
Six Principles of Voice Design: Designing for the Ear and Encoding It into Agent Script
Salesforce Blog · 8 oct. 2026