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.
Academic integrity is honest, fair, respectful, and responsible behaviour in an academic environment. When considering the use of any LLM/GenAI tool you first must remember:
As a next step towards upholding your academic integrity in relation to GenAI, it is essential to understand the implications of misinformation, lack of transparency and bias.
Because GenAI is trained on real-world data, text, and media from the internet, the content it provides may be misleading, factually inaccurate, or outright misinformation such as deep fakes. The reliability of a source is not given consideration by a LLM which means a peer-reviewed academic article and a Reddit post are considered of equal authority. As such, the output of a LLM may not always be credible or reliable and can reflect implicit or explicit biases, outdated information, or fabricated content (TLP, 2025).
Concerns about the sources used to train the data are compounded by the fact that GenAI tools are often unable to replicate outputs and unable to correctly and consistently cite specific references. This is problematic particularly in academic contexts where your assignments require citations to uphold academic standards. As such, you must always verify the accuracy of any GenAI-generated content by using other reliable sources before including it in your work. Once you have verified content, you must also properly cite GenAI including the prompts used as inputs.
Most LLMs are designed to benefit the people who already possess the most power and privilege in the world. Their design and development has not prioritized ethical engagement with historically marginalized communities and as such GenAI / LLMs are known to reproduce this ongoing bias and exclusion (Sweetman, 2024).
“Poet of Code shares "AI, Ain't I A Woman" - a spoken word piece that highlights how artificial intelligence can misinterpret the images of iconic black women: Oprah, Serena Williams, Michelle Obama, Sojourner Truth, Ida B. Wells, and Shirley Chisholm” (Buolamwini, 2018).
A recent UNESCO report (2024) identifies gender bias in GenAI systems as a global, persistent issue that serves to reinforce, “perpetuate (and even scale and amplify) human, structural and social biases. These biases not only prove difficult to mitigate, but may also lead to harm at the individual, collective, or societal level” (UNESCO, p.3).
One study of gendered names showed a strong, “deep-seated bias in how LLMs represent gender in relation to careers” and family roles, “where female names were associated with “home,” “family,” “children,” and “marriage”; while male names were associated with “business,” “executive,” “salary,” and “career” (UNESCO, p.9).”
This bias reproduction is further reflected in how GenAI tools currently assess, select, exclude and make recommendations. In a Scientific American study (2023), researchers asked LLMs to produce recommendation letters for hypothetical employees and “observed significant gender biases,” where “ChatGPT deployed nouns such as “expert” and “integrity” for men”” and was more likely to call women beautiful and delightful (Stokel-Walker, 2023).