Language Generation Models Can Cause Harm: So What Can We Do About It? An Actionable Survey

By Sachin Kumar, Vidhisha Balachandran, Lucille Njoo, Antonios Anastasopoulos, Yulia Tsvetkov. 2022

119 citations recorded by the source corpus.

Recent advances in the capacity of large language models to generate human-like text have resulted in their increased adoption in user-facing settings. In parallel, these improvements have prompted a heated discourse around the risks of societal harms they introduce, whether inadvertent or malicious. Several studies have explored these harms and called for their mitigation via development of safer, fairer models. Going beyond enumerating the risks of harms, this work provides a survey of practical methods for addressing potential threats and societal harms from language generation models. We draw on several prior works’ taxonomies of language model risks to present a structured overview of strategies for detecting and ameliorating different kinds of risks/harms of language generators. Bridging diverse strands of research, this survey aims to serve as a practical guide for both LM researchers and practitioners, with explanations of different strategies’ motivations, their limitations, and open problems for future research.

Subfields: Natural Language Processing, Safety, Robustness & Interpretability

How researchers see this paper

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

In Natural Language Processing

In Safety, Robustness & Interpretability

Citations in the corpus

Llama 2: Open Foundation and Fine-Tuned Chat Models (2023, background): “Solaiman et al. (2023) categorizes these impacts into two groups — those that can be assessed within the base system and those requiring a societal context evaluation, while Kumar et al. (2022) offers potential mitigation strategies to curb harm.”
AI Transparency in the Age of LLMs: A Human-Centered Research Roadmap (2023, background): “…contributing to automation bias, overreliance, or automation-induced complacency (Parasuraman and Manzey, 2010; Wickens et al., 2015). p While the capabilities of LLMs are impressive, they also raise new risks (Bender et al., 2021; Kumar et al., 2023; OpenAI, 2023; Weidinger et al., 2022).”
A framework to assess clinical safety and hallucination rates of LLMs for medical text summarisation (2025, unknown): “The problem of hallucinations poses a signi ﬁ cant challenge to date 1,13 .”
Bias and Fairness in Large Language Models: A Survey (2023, background): “Kumar et al. (2023b) also warn that decoding algorithms may be manipulated to generate biased language.”
From Pretraining Data to Language Models to Downstream Tasks: Tracking the Trails of Political Biases Leading to Unfair NLP Models (2023, methodology): “NLP models (Blodgett et al., 2020a; Kumar et al., 2022), designed synthetic datasets (Nangia et al.”
Gradient-based Constrained Sampling from Language Models (2022, background): “future research should focus on developing better defense methods against misusing these models, in a way that could cause societal harms (Kumar et al., 2022).”