Angjelin Hila

Angjelin Hila

This paper takes an ecological approach toward large-scale models of hybrid human-AI intelligence. Emerging models of human-AI interaction predominantly advance the complementarity thesis variously dubbed human-AI collaboration and human-AI hybrid intelligence. However, this constitutes an over-simplification of the modalities of human-AI interaction and possibility-space for both individual and collective action that human-AI interaction potentiates. To fill these gaps, this paper develops a decision and game-theoretic approach to the human-AI delegation-verification dilemma. First, we map out canonical decision-theoretic strategies that account for adaptive user trajectories, modeling how agents transition between strategies based on interaction feedback to reach stable equilibria. Second, we scale individually stable strategies to collective equilibria using three extrapolation principles: (a) non-communicative aggregation (b) local social signaling and (c) institutional norms setting. The analysis identifies the emergence of sociotechnical lock-in, a macro-behavioral state where individually adaptive delegation, in the absence of communicative and institutional safeguards, aggregates into a systemic collective action problem modeled as a prisoner's dilemma that degrades shared epistemic standards. We argue that adoption under higher communicative standards and institutional norms can mitigate suboptimal collective equilibria by imposing social commitments on individual users.

Angjelin Hila

We examine epistemological threats posed by human and LLM interaction. We develop collective epistemology as a theory of epistemic warrant distributed across human collectives, using bounded rationality and dual process theory as background. We distinguish internalist justification, defined as reflective understanding of why a proposition is true, from externalist justification, defined as reliable transmission of truths. Both are necessary for collective rationality, but only internalist justification produces reflective knowledge. We specify reflective knowledge as follows: agents understand the evaluative basis of a claim, when that basis is unavailable agents consistently assess the reliability of truth sources, and agents have a duty to apply these standards within their domains of competence. We argue that LLMs approximate externalist reliabilism because they can reliably transmit information whose justificatory basis is established elsewhere, but they do not themselves possess reflective justification. Widespread outsourcing of reflective work to reliable LLM outputs can weaken reflective standards of justification, disincentivize comprehension, and reduce agents' capacity to meet professional and civic epistemic duties. To mitigate these risks, we propose a three tier norm program that includes an epistemic interaction model for individual use, institutional and organizational frameworks that seed and enforce norms for epistemically optimal outcomes, and deontic constraints at organizational and or legislative levels that instantiate discursive norms and curb epistemic vices.

Angjelin Hila, Elliott Hauser

In this study, we investigate the use of large language models (LLMs), specifically ChatGPT, for structured deductive qualitative coding. While most current research emphasizes inductive coding applications, we address the underexplored potential of LLMs to perform deductive classification tasks aligned with established human-coded schemes. Using the Comparative Agendas Project (CAP) Master Codebook, we classified U.S. Supreme Court case summaries into 21 major policy domains. We tested four intervention methods: zero-shot, few-shot, definition-based, and a novel Step-by-Step Task Decomposition strategy, across repeated samples. Performance was evaluated using standard classification metrics (accuracy, F1-score, Cohen's kappa, Krippendorff's alpha), and construct validity was assessed using chi-squared tests and Cramer's V. Chi-squared and effect size analyses confirmed that intervention strategies significantly influenced classification behavior, with Cramer's V values ranging from 0.359 to 0.613, indicating moderate to strong shifts in classification patterns. The Step-by-Step Task Decomposition strategy achieved the strongest reliability (accuracy = 0.775, kappa = 0.744, alpha = 0.746), achieving thresholds for substantial agreement. Despite the semantic ambiguity within case summaries, ChatGPT displayed stable agreement across samples, including high F1 scores in low-support subclasses. These findings demonstrate that with targeted, custom-tailored interventions, LLMs can achieve reliability levels suitable for integration into rigorous qualitative coding workflows.