Large-scale Pre-Trained Language Models (PTLMs) capture knowledge from massive human-written data which contains latent societal biases and toxic contents. In this paper, we leverage the primary task of PTLMs, i.e., language modeling, and propose a new metric to quantify manifested implicit representational harms in PTLMs towards 13 marginalized demographics. Using this metric, we conducted an empirical analysis of 24 widely used PTLMs. Our analysis provides insights into the correlation between the proposed metric in this work and other related metrics for representational harm. We observe that our metric correlates with most of the gender-specific metrics in the literature. Through extensive experiments, we explore the connections between PTLMs architectures and representational harms across two dimensions: depth and width of the networks. We found that prioritizing depth over width, mitigates representational harms in some PTLMs. Our code and data can be found at https://github.com/microsoft/SafeNLP.
Beyond Perplexity: Multi-dimensional Safety Evaluation of LLM Compression
Kahlert School of Computing, University of Utah（犹他大学卡勒特计算学院）；Google DeepMind（谷歌DeepMind）；Scientific Computing and Imaging Institute, University of Utah（犹他大学科学计算与成像研究所）
As machine learning methods are deployed in real-world settings such as healthcare, legal systems, and social science, it is crucial to recognize how they shape social biases and stereotypes in these sensitive decision-making processes. Among such real-world deployments are large-scale pretrained language models (LMs) that can be potentially dangerous in manifesting undesirable representational biases - harmful biases resulting from stereotyping that propagate negative generalizations involving gender, race, religion, and other social constructs. As a step towards improving the fairness of LMs…
Many educational technologies use artificial intelligence (AI) that presents generated or produced language to the learner. We contend that all language, including all AI communication, encodes information about the identity of the human or humans who contributed to crafting the language. With AI communication, however, the user may index identity information that does not match the source. This can lead to representational harms if language associated with one cultural group is presented as "standard" or "neutral", if the language advantages one group over another, or if the language reinforc…
KU Leuven（鲁汶大学）；Leuven.AI institute（鲁汶人工智能研究所）；TU Berlin（柏林工业大学）；Weizenbaum Institute（魏泽鲍姆研究所）
Large pre-trained language models are successfully being used in a variety of tasks, across many languages. With this ever-increasing usage, the risk of harmful side effects also rises, for example by reproducing and reinforcing stereotypes. However, detecting and mitigating these harms is difficult to do in general and becomes computationally expensive when tackling multiple languages or when considering different biases. To address this, we present FairDistillation: a cross-lingual method based on knowledge distillation to construct smaller language models while controlling for specific bias…
University of Southern California（南加州大学）；Information Sciences Institute（信息科学研究所）
Warning: this paper contains content that may be offensive or upsetting. Numerous natural language processing models have tried injecting commonsense by using the ConceptNet knowledge base to improve performance on different tasks. ConceptNet, however, is mostly crowdsourced from humans and may reflect human biases such as "lawyers are dishonest." It is important that these biases are not conflated with the notion of commonsense. We study this missing yet important problem by first defining and quantifying biases in ConceptNet as two types of representational harms: overgeneralization of polar…
Do Large Language Models Advocate for Inferentialism?
College of Media Arts, Science and Technology, School of Informatics, University of Tsukuba（筑波大学媒体艺术、科学与技术学院，信息学院）；Institute of Systems and Information Engineering, University of Tsukuba（筑波大学系统与信息工程研究所）
