AI brain balancing on a tightrope

Are AI's Making Risky Choices? Unveiling How LLMs Decide Under Pressure

Nico Varela in Tech & Innovation December 2025 • 4 min read.

"A new framework evaluates large language models (LLMs) to see if their decision-making aligns with ethical expectations or harbors hidden biases."

Large language models (LLMs) are increasingly used to support crucial decision-making across various fields. From healthcare to finance, these AI systems provide sophisticated responses and assist in complex processes. But how do LLMs truly handle risk and uncertainty? Do their decision-making tendencies align with human norms, or do they exhibit hidden biases?

A groundbreaking new study from the University of Illinois at Urbana-Champaign introduces a comprehensive framework for evaluating the decision-making behaviors of LLMs. This framework, grounded in behavioral economics theories, assesses LLMs across three key dimensions: risk preference, probability weighting, and loss aversion. By understanding these aspects, we can better determine whether LLMs are making sound, ethical choices.

The study dives deep into the internal decision-making processes of LLMs, examining their behavior in both context-free settings and when embedded with socio-demographic features. This research uncovers critical insights into the potential biases and ethical considerations that arise when deploying LLMs in real-world scenarios. Are LLMs truly objective, or do they carry the weight of societal prejudices?

Decoding the AI Mind: How LLMs Weigh Risk and Uncertainty

The research framework is built upon established behavioral economics theories, particularly the value function model proposed by Tanaka, Camerer, and Nguyen (TCN model). This model enables the evaluation of risk preferences (how willing an LLM is to take chances), probability weighting (how an LLM perceives the likelihood of different outcomes), and loss aversion (how strongly an LLM avoids potential losses).

The study utilized a multiple-choice-list experiment to gauge the decision-making tendencies of three prominent commercial LLMs: ChatGPT-4.0-Turbo, Claude-3-Opus, and Gemini-1.0-pro. In a context-free setting, these models were presented with various scenarios involving potential gains and losses, allowing researchers to analyze their inherent preferences.

Risk Aversion: LLMs generally exhibit risk-averse behavior, similar to humans, preferring to avoid potential losses.
Loss Aversion: LLMs demonstrate a tendency to avoid losses more strongly than they seek gains.
Probability Weighting: LLMs tend to overweight small probabilities, meaning they might overestimate the likelihood of rare events.
Model Variations: Significant variations exist in the degree to which these behaviors are expressed across different LLMs.

Interestingly, while all three models showed risk-averse tendencies, their approaches differed. ChatGPT leaned towards conservative choices, while Claude adopted a riskier approach with higher loss aversion. Gemini balanced risk and caution, exhibiting moderate risk-taking tendencies.

Ethical AI: Charting a Course for Responsible Decision-Making

This research highlights the critical need for ongoing scrutiny and refinement of LLMs to ensure they do not perpetuate or exacerbate societal biases. By understanding how LLMs make decisions and identifying potential biases, we can work towards developing standards and guidelines for ethical AI deployment. As LLMs become further integrated into our lives, it is our responsibility to ensure they operate within ethical boundaries, promoting fairness and equity for all.

About this Article -

This article was crafted using a human-AI hybrid and collaborative approach. AI assisted our team with initial drafting, research insights, identifying key questions, and image generation. Our human editors guided topic selection, defined the angle, structured the content, ensured factual accuracy and relevance, refined the tone, and conducted thorough editing to deliver helpful, high-quality information.See our About page for more information.

This article is based on research published under:

DOI-LINK: https://doi.org/10.48550/arXiv.2406.05972,

Title: Decision-Making Behavior Evaluation Framework For Llms Under Uncertain Context

Subject: cs.ai cs.cy cs.hc cs.lg econ.th

Authors: Jingru Jia, Zehua Yuan, Junhao Pan, Paul E. Mcnamara, Deming Chen

Published: 09-06-2024

Everything You Need To Know

What are the key dimensions used to evaluate the decision-making of LLMs within the new framework?

The framework evaluates Large Language Models (LLMs) based on three key dimensions: risk preference, which assesses the willingness of the LLM to take risks; probability weighting, which examines how the LLM perceives the likelihood of different outcomes; and loss aversion, which measures how strongly the LLM avoids potential losses. Understanding these dimensions allows us to assess whether LLMs are making sound and ethical choices, similar to human decision-making processes.

How do LLMs like ChatGPT, Claude, and Gemini differ in their approaches to risk, according to the study?

The study revealed variations in how different LLMs approach risk. ChatGPT exhibited conservative tendencies, favoring less risky options. Claude took a riskier approach, with higher loss aversion, indicating a stronger avoidance of potential losses. Gemini, on the other hand, balanced risk and caution, showing moderate risk-taking behaviors.

What is the significance of probability weighting in the context of LLMs?

Probability weighting is crucial because it reveals how an LLM perceives and processes the likelihood of different outcomes. LLMs tend to overweight small probabilities, which means they might overestimate the chance of rare events. This can lead to decisions that are disproportionately influenced by low-probability scenarios, potentially resulting in unexpected or undesirable outcomes, highlighting the need for careful consideration in high-stakes applications.

How does the research framework assess the decision-making behaviors of LLMs?

The research framework is built upon behavioral economics theories, particularly the value function model proposed by Tanaka, Camerer, and Nguyen (TCN model). This model enables the evaluation of risk preferences (how willing an LLM is to take chances), probability weighting (how an LLM perceives the likelihood of different outcomes), and loss aversion (how strongly an LLM avoids potential losses). The study utilizes a multiple-choice-list experiment to gauge the decision-making tendencies of three prominent commercial LLMs: ChatGPT-4.0-Turbo, Claude-3-Opus, and Gemini-1.0-pro. In a context-free setting, these models were presented with various scenarios involving potential gains and losses, allowing researchers to analyze their inherent preferences.

Why is it important to understand the ethical implications of how LLMs make decisions?

Understanding the ethical implications of LLM decision-making is crucial because these models are increasingly used in critical areas like healthcare and finance. If LLMs exhibit hidden biases or make decisions that don't align with ethical expectations, it can lead to unfair or discriminatory outcomes. By understanding risk preferences, probability weighting, and loss aversion, we can work towards developing standards and guidelines for ethical AI deployment, ensuring fairness, equity, and responsible use of these powerful technologies. It is our responsibility to ensure that Large Language Models operate within ethical boundaries, promoting fairness and equity for all as LLMs are integrated into our lives.