Always verify AI-generated information and any sources it produces.
AI tools may:
generate inaccurate, oversimplified, or exaggerated information and summaries
fabricate citations or sources (“hallucinations”)
recommend promotional or low-quality websites as reliable sources
Consultation with the GB Library's guide on Responsible Use of Generative AI is recommended to understand key terms, how to evaluate outputs, academic integrity, harm considerations, and copyright.
Please note uploading library e-resources to GenAI tools is not permitted and may constitute copyright infringement. Please refer to our Copyright Best Practices for more details.
"Developing critical GenAI literacy means confronting algorithmic bias, promoting diverse perspectives, and embedding justice, equity, and critical thinking into the learning process" (Rapanta et al., 2025). A first step towards critical literacy is understanding some harm considerations of LLMs.
Citation: Sweetman, Rebecca. “Some Harm Considerations of Large Language Models (LLMs).” Created March 2023, last updated August 2024. License: Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).
The development, training and use of LLMs require systems that use a significant amount of energy, contribute to carbon emissions, consume vast amounts of water for cooling, and produce electronic waste (TLP, 2025). If you value and prioritize sustainability, it is crucial to consider whether the use of GenAI tools is worth its proven negative environmental impacts.
GenAI & LLMs enhance the probability for the automation of a wide spectrum of lower wage work, including customer service. This harm may cause increased precarity and unemployment for the world’s most underemployed and marginalized workers who are already disproportionately employed in lower wage work (Sweetman, 2024).
This academic article identified several of the negative effects that LLMs have on employment, including: “increasing inequality and negative effects on job quality,” “undermining creative economies,” and “displacing employees from their roles” leading to “an increase in unemployment” (Weidinger et al., 2021).
The economic impact on creative communities and industries “extends far beyond individual creators. Independent artists and small creative enterprises find their market share diminishing as platform-mediated distribution favours content optimized for algorithmic promotion, often at the expense of originality and diversity” (CULTAI, p.35).
The impact on social justice, worldwide and particularly in the Global South is significant. GenAI and LLMs not only reproduce biased datasets, overemphasize dominant culture, amplify toxic cultural norms, and normalize privilege bias, they are designed to do exactly that. This will only serve to perpetuate ableist, ecocidal, gendered, genocidal, racist, classist and socioeconomic harms as biased datasets increasingly reaffirm themselves (Sweetman, 2024).