TrustLLM: Trustworthiness in Large Language Models

By Lichao Sun, Yue Huang, Haoran Wang, Siyuan Wu, Qihui Zhang, Chujie Gao, Yixin Huang, Wenhan Lyu, Yixuan Zhang, Xiner Li and 56 more. arXiv.org, 2024

353 citations recorded by the source corpus.

Large language models (LLMs), exemplified by ChatGPT, have gained considerable attention for their excellent natural language processing capabilities. Nonetheless, these LLMs present many challenges, particularly in the realm of trustworthiness. Therefore, ensuring the trustworthiness of LLMs emerges as an important topic. This paper introduces TrustLLM, a comprehensive study of trustworthiness in LLMs, including principles for different dimensions of trustworthiness, established benchmark, evaluation, and analysis of trustworthiness for mainstream LLMs, and discussion of open challenges and future directions. Specifically, we first propose a set of principles for trustworthy LLMs that span eight different dimensions. Based on these principles, we further establish a benchmark across six dimensions including truthfulness, safety, fairness, robustness, privacy, and machine ethics. We then present a study evaluating 16 mainstream LLMs in TrustLLM, consisting of over 30 datasets. Our findings firstly show that in general trustworthiness and utility (i.e., functional effectiveness) are positively related. Secondly, our observations reveal that proprietary LLMs generally outperform most open-source counterparts in terms of trustworthiness, raising concerns about the potential risks of widely accessible open-source LLMs. However, a few open-source LLMs come very close to proprietary ones. Thirdly, it is important to note that some LLMs may be overly calibrated towards exhibiting trustworthiness, to the extent that they compromise their utility by mistakenly treating benign prompts as harmful and consequently not responding. Finally, we emphasize the importance of ensuring transparency not only in the models themselves but also in the technologies that underpin trustworthiness. Knowing the specific trustworthy technologies that have been employed is crucial for analyzing their effectiveness.

Subfields: Artificial Intelligence, Health Informatics

View Paper: https://www.semanticscholar.org/paper/fb4dc0178e5d7347b1615c48caf05347b6e5eb48

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Health Informatics

Citations in the corpus

Large Language Models: A Survey (2024, neutral): “Ensuring the robustness and security of LLMs against adversarial attacks and other vulnerabilities is a critical area of research [219].”
TÜLU 3: Pushing Frontiers in Open Language Model Post-Training (2024, neutral): “JailbreakTrigger [Huang et al., 2024b] incorporates prompts based on 13 distinct jailbreak attack methods.”
A Survey on Knowledge Distillation of Large Language Models (2024, methodology): “Recently, distillation-based methods are also proposed (Schick et al., 2023; Zhang, 2023; Patil et al., 2023; Tang et al., 2023a; Qin et al., 2023a; Yuan et al., 2023a; Gao et al., 2023b; Wang et al., 2024; Shen et al., 2024; Yuan et al., 2024b).”
Sora: A Review on Background, Technology, Limitations, and Opportunities of Large Vision Models (2024, background): “One primary area of focus is the model’s safety, specifically its resilience against misuse and so-called “jailbreak” attacks, where users attempt to exploit vulnerabilities to generate prohibited or harmful content [107, 108, 109, 110, 111, 112, 113, 114, 115, 116].”
Understanding the Effects of Iterative Prompting on Truthfulness (2024, background): “Moreover, the introduction of benchmarks for evaluating model trustworthiness, including aspects of truthfulness, provides a comprehensive framework for assessing and comparing the performance of various LLMs, paving the way for more trustworthy AI systems [21].”
JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Models (2024, background): “Moreover, we strive to make our benchmark community-driven, prioritizing clear guidelines for adding new attacks, models, and defenses.”
Empowering Biomedical Discovery with AI Agents (2024, neutral): “An AI agent should be able to formulate biomedical hypotheses, critically evaluate them, characterize their uncertainty, and use that as a driver to acquire and reﬁne its scientiﬁc knowledge bases in a way that human scientists can trust [17].”