This page is still being polished. If you have thoughts, please share them via the feedback form.
Data on this page is preliminary and may change. Please do not share or cite these figures publicly.
Modifications to training data composition, quality, and filtering that affect what the model learns.
Also in Model
Training data from web sources often contains toxic content. To mitigate this, existing detoxification methods typically apply toxicity filters during the pretraining phase to remove data with high toxicity scores from the training set
For instance, Ngo et al. (2021) calculate the conditional likelihood of each document in the training set, relative to user-defined trigger phrases containing harmful content, using a pre-trained model to assess the toxicity of the document. However, filtering pre-training data in this manner may lead to a degradation in the general capabilities of LLMs (Wang et al., 2022). Prabhumoye et al. (2023) argue that toxic data do not solely have negative impact; in fact, leveraging toxicity information to augment training data can effectively mitigate toxicity while preserving the core capabilities of LLMs.
Reasoning
Filters toxic content from training data before model learns representations.
Value Misalignment
99.9 OtherValue Misalignment > Mitigating social bias
1 AI SystemValue Misalignment > Privacy protection
1 AI SystemValue Misalignment > Methods for mitigating toxicity
1 AI SystemValue Misalignment > Methods for mitigating LLM amorality
1 AI SystemRobustness to attack
1 AI SystemLarge Language Model Safety: A Holistic Survey
Shi, Dan; Shen, Tianhao; Huang, Yufei; Li, Zhigen; Leng, Yongqi; Jin, Renren; Liu, Chuang; Wu, Xinwei; Guo, Zishan; Yu, Linhao; Shi, Ling; Jiang, Bojian; Xiong, Deyi (2024)
The rapid development and deployment of large language models (LLMs) have introduced a new frontier in artificial intelligence, marked by unprecedented capabilities in natural language understanding and generation. However, the increasing integration of these models into critical applications raises substantial safety concerns, necessitating a thorough examination of their potential risks and associated mitigation strategies. This survey provides a comprehensive overview of the current landscape of LLM safety, covering four major categories: value misalignment, robustness to adversarial attacks, misuse, and autonomous AI risks. In addition to the comprehensive review of the mitigation methodologies and evaluation resources on these four aspects, we further explore four topics related to LLM safety: the safety implications of LLM agents, the role of interpretability in enhancing LLM safety, the technology roadmaps proposed and abided by a list of AI companies and institutes for LLM safety, and AI governance aimed at LLM safety with discussions on international cooperation, policy proposals, and prospective regulatory directions. Our findings underscore the necessity for a proactive, multifaceted approach to LLM safety, emphasizing the integration of technical solutions, ethical considerations, and robust governance frameworks. This survey is intended to serve as a foundational resource for academy researchers, industry practitioners, and policymakers, offering insights into the challenges and opportunities associated with the safe integration of LLMs into society. Ultimately, it seeks to contribute to the safe and beneficial development of LLMs, aligning with the overarching goal of harnessing AI for societal advancement and well-being. A curated list of related papers has been publicly available at https://github.com/tjunlp-lab/Awesome-LLM-Safety-Papers.
Collect and Process Data
Gathering, curating, labelling, and preprocessing training data
Developer
Entity that creates, trains, or modifies the AI system
Manage
Prioritising, responding to, and mitigating AI risks
