Large language models (LLMs) are increasingly used to simulate human behavior in social settings such as legal mediation, negotiation, and dispute resolution. However, it remains unclear whether these simulations reproduce the personality–behavior patterns observed in humans. Human personality, for instance, shapes how individuals navigate social interactions, including strategic choices and behaviors in emotionally charged interactions. This raises the question: Can LLMs, when prompted with personality traits, reproduce personality-driven differences in human conflict behavior? To explore this, we introduce an evaluation framework that enables direct comparison of human-human and LLM-LLM behaviors in dispute resolution dialogues with respect to Big Five Inventory (BFI) personality traits. This framework provides a set of interpretable metrics related to strategic behavior and conflict outcomes. We additionally contribute a novel dataset creation methodology for LLM dispute resolution dialogues with matched scenarios and personality traits with respect to human conversations. Finally, we demonstrate the use of our evaluation framework with three contemporary closed-source LLMs and show significant divergences in how personality manifests in conflict across different LLMs compared to human data, challenging the assumption that personality-prompted agents can serve as reliable behavioral proxies in socially impactful applications. Our work highlights the need for psychological grounding and validation in AI simulations before real-world use.

James Anthony Hale, Sushrita Rakshit, Kushal Chawla, Jeanne M Brett, Jonathan Gratch. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Abstract Prior work suggests automated dispute resolution tools using “provably fair” algorithms can address disparities between demographic groups. These methods use multi-criteria elicited preferences from all disputants and satisfy constraints to generate “fair” solutions. However, we analyze the potential for inequity to permeate proposals through the preference elicitation stage. This possibility arises if differences in dispositional attitudes differ between demographics, and those dispositions affect elicited preferences. Specifically, risk aversion plays a prominent role in predicting preferences. Risk aversion predicts a weaker relative preference for salary and a softer within-issue utility for each issue; this leads to worse compensation packages for risk-averse groups. These results raise important questions in AI-value alignment about whether an AI mediator should take explicit preferences at face value.

The intelligent virtual agent community often works from the assumption that embodiment confers clear benefits to human-machine interaction. However, embodiment has potential drawbacks in highlighting the salience of social stereotypes such as those around race and gender. Indeed, theories of computer-mediated communication highlight that visual anonymity can sometimes enhance team outcomes. Negotiation is one domain where social perceptions can impact outcomes. For example, research suggests women perform worse in negotiations and find them more aversive, particularly when interacting with men opponents. Research with human participants makes it challenging to unpack whether these negative consequences stem from women’s perceptions of their partner or greater toughness on the part of these men opponents. We use a socially intelligent AI negotiation agent to begin to unpack these processes. We manipulate the perceived toughness of the AI by whether or not it expresses anger — a common tactic to extract concessions. Independently, we manipulate the activation of stereotypes by randomly setting whether the interaction has embodiment (as a male opponent) or has only text (where we obscure gender cues). We find a clear interaction between gender and embodiment. Specifically, women perform worse, and men perform better against an apparently male opponent compared to a disembodied agent – as measured by the subjective value they assign to their outcome. This highlights the potential disadvantages of embodiment in negotiation, though future research must rule out alternative mechanisms that might explain these results.

Current approaches to building negotiation agents rely either on model-based techniques that explicitly implement key principles of negotiation or model-free techniques leveraging algorithms developed via training on large amounts of human-generated text. We bridge these two approaches by combining a model-based approach with large language models for natural language understanding and generation. We find large language models perform well at recognizing dialogue acts and an opponent's emotions; perform reasonably well at recognizing opponents' preferences in the negotiation; and perform worse at understanding opponent offers. We also perform a qualitative comparison of the capabilities of our hybrid approach with a model-free method and find our hybrid agent provides safeguards against hallucinations and guarantees more control over aspects of negotiation such as emotional expressions, information sharing, and concession strategies.

Much prior work in automated negotiation makes the simplifying assumption of linear utility functions. As such, we propose a framework for multilateral repeated negotiations in a complex game setting---to introduce non-linearities---where negotiators can choose with whom they negotiate in subsequent games. This game setting not only creates non-linear utility functions, but also motivates the negotiation.

The authors address the problem of enhancing hybrid magnetic resonance (MR) images degraded by T2 effects and additive measurement noise. To reduce imaging time, MR signals are acquired using hybrid imaging (HI) sequences such as rapid acquisition relaxation-enhanced (RARE) and fast spin-echo (FSE). With these techniques, T2 effects act as a distortion filter. This T2 filter affects the signal and results in image spatial resolution and/or contrast loss. Furthermore, the amplitude and phase discontinuities in the T2 filter frequency response function may generate serious ringing artifacts. These distortions will damage image quality and affect object detectability. The authors use the Wiener filter and linear prediction (LP) technique to process HI MR signals in the spatial frequency domain (K-space) and the hybrid domain, respectively. Based on the average amplitude symmetry constraint of the spin echo signal, the amplitude frequency response function of the T2 distortion filter can be estimated and used in the Wiener filter for a global T2 amplitude restoration. Then, the linear prediction technique is utilized to obtain the local signal amplitude and phase estimates around the discontinuities of the frequency response function of the T2 filter. These estimates are used to make local amplitude and phase corrections. The effectiveness of this combined technique in correcting T2 distortion and reducing the measurement noise is analyzed and demonstrated using experiments on both phantoms and human studies.

A filtering algorithm for fast image enhancement is described. The algorithm tries to make the minimum modification on the original image structures while it performs noise smoothing at a given filtering level. The filtered image is a weighted combination of four subimages obtained from low-pass filtering the original image along four major directions. The weighting on each subimage is controlled by the differences between these subimages and the original image The resulting image is then nonsymmetrically sharpened to enhance the image structure boundaries, The overall effect of this filtering structure is effective adaptive noise reduction and edge enhancement with an efficient implementation using array processors. The high regularity and parallelism of the algorithm also makes it suitable for its efficient implementation using very large scale integrated (VLSI) circuits or multiprocessor systems. The performance of the algorithm in effectively reducing image noise and preserving/enhancing important image structures is discussed and demonstrated using several MR images from a low-field-strength MR imaging system.