TrustLLM: Trustworthiness in Large Language Models
TrustLLM: Надёжность больших языковых моделей
2024-01-01
SCID: 54.1/25rmddwz
Discuss with AI
LLM trustworthinessTrustLLM benchmarkfairness and robustnessprivacy and machine ethicstruthfulness and safety
Figures from the paper
Abstract (AI)
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.
Key Findings
1
Evaluation of 16 mainstream LLMs finds that trustworthiness and functional utility are generally positively related.
2
Proprietary LLMs generally outperform most open-source models in trustworthiness, although several open-source models approach proprietary performance.
3
Some models over-calibrate toward trustworthiness, incorrectly treating benign prompts as harmful and reducing utility by refusing to respond.
4
The benchmark covers six trustworthiness dimensions—truthfulness, safety, fairness, robustness, privacy, and machine ethics—and includes over 30 datasets.
5
The study emphasizes transparency about both model behavior and the specific technologies used to achieve trustworthiness.
6
TrustLLM introduces principles for trustworthy large language models spanning eight dimensions and uses them to structure systematic evaluation.
Research Object
large language models (LLMs)
Research Subject
trustworthiness of LLMs across truthfulness, safety, fairness, robustness, privacy, and machine ethics, including its relationship with utility
Publication Details
Publication Date
2024-01-01
Journal
Publisher
ISSN
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest