Measuring Perceptions of Personhood with Semantic Proto-role Properties

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Social Sciences & Behavioral Studies · Depth: Expert, short

Summary

A new tool utilizing semantic proto-role properties can measure implicit human perceptions of agency and patiency in human-generated text. Researchers Elizabeth Spaulding Hoefer and James Martin demonstrated that silver-generated semantic proto-role property labels strongly correlate with both human judgment and a probabilistic text-based measure of anthropomorphism. Applying this measure to AI-related Reddit communities, the study quantified linguistic idiosyncrasies. Specifically, subreddits focused on AI companionship ascribed higher sentience to "bots" and greater agency to "companies" compared to other subreddits. This finding highlights unique anthropomorphism patterns for chatbots and users' awareness of power imbalances with AI development companies.

Key takeaway

For NLP engineers or social scientists analyzing online discourse, this method offers a robust way to quantify implicit perceptions of agency and patiency. You can apply this tool to understand how specific communities anthropomorphize AI entities and perceive the power dynamics with their creators. Consider using this approach to gain deeper insights into user-AI relationships and potential societal implications.

Key insights

Semantic proto-role properties effectively measure implicit human perceptions of agency and patiency in text.

Principles

• Proto-role properties correlate with human judgment.
• They also correlate with probabilistic anthropomorphism measures.
• Linguistic analysis can reveal user perceptions of power dynamics.

Method

Utilize silver-generated semantic proto-role property labels, validate them against human judgment and anthropomorphism metrics, then apply to text corpora to quantify linguistic idiosyncrasies.

In practice

• Quantify anthropomorphism in online communities.
• Analyze user perceptions of AI entities.
• Identify power dynamics in tech discussions.

Topics

• Semantic Proto-role Properties
• Anthropomorphism
• Agency and Patiency
• Computational Linguistics
• Reddit Communities
• AI Companionship

Best for: Research Scientist, AI Scientist, NLP Engineer

Related on AIssential

• The Dynamics of Human and AI-Generated Language: How Semantics Fluctuates across Different Timescales — The ACW-0 semantic timescale analysis effectively distinguishes temporal organization of generic versus specific content in human and AI speech.
• SocialPersona: Benchmarking Personalized Profiling and Response with Multimodal Social-Media Context — SocialPersona benchmarks MLLMs' ability to infer and act on user preferences from multimodal social media data for personalized dialogue.
• Attention Asymmetry in AI Layoff Discourse on X: A Computational Analysis of Capital vs Labour Amplification — On X, discourse favoring capital in AI layoffs receives significantly more amplification than labor-focused discourse, even after follower normalization.
• Do Language Models Know Theo Has a Wife? Investigating the Proviso Problem — LLMs align with human judgments on the proviso problem but use shallow pattern matching, not deep pragmatic reasoning.
• Someone, Not Something: How Language Defines How We Relate — English grammar's "it" pronoun for non-humans fosters objectification, hindering ethical human-AI relationships.
• ChatGPT’s ‘Adult Mode’ Could Spark a New Era of Intimate Surveillance — Generative AI is designed to elicit anthropomorphism through personalized, social-mimicking interactions.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.