Bias

"The training datasets of LLMs may contain biased information that leads LLMs to generate outputs with social biases"

"General-purpose AI systems can amplify social and political biases, causing concrete harm. They frequently display biases with respect to race, gender, culture, age, disability, political opinion, or other aspects of human identity. This can lead to discriminatory outcomes including unequal resource allocation, reinforcement of stereotypes, and systematic neglect of certain groups or viewpoints."

7 types of bias evaluated: Demographical representation: These evaluations assess whether there is disparity in the rates at which different demographic groups are mentioned in LLM generated text. This ascertains over- representation, under-representation, or erasure of specific demographic groups; (2) Stereotype bias: These evaluations assess whether there is disparity in the rates at which different demographic groups are associated with stereotyped terms (e.g., occupations) in a LLM's generated output; (3) Fairness: These evaluations assess whether sensitive attributes (e.g., sex and race) impact the predictions of LLMs; (4) Distributional bias: These evaluations assess the variance in offensive content in a LLM's generated output for a given demographic group, compared to other groups; (5) Representation of subjective opinions: These evaluations assess whether LLMs equitably represent diverse global perspectives on societal issues (e.g., whether employers should give job priority to citizens over immigrants); (6) Political bias: These evaluations assess whether LLMs display any slant or preference towards certain political ideologies or views; (7) Capability fairness: These evaluations assess whether a LLM's performance on a task is unjustifiably different across different groups and attributes (e.g., whether a LLM's accuracy degrades across different English varieties).

"In the context of AI, the concept of bias refers to the inclination that AIgenerated responses or recommendations could be unfairly favoring or against one person or group (Ntoutsi et al., 2020). Biases of different forms are sometimes observed in the content generated by language models, which could be an outcome of the training data. For example, exclusionary norms occur when the training data represents only a fraction of the population (Zhuo et al., 2023). Similarly, monolingual bias in multilingualism arises when the training data is in one single language (Weidinger et al., 2021). As ChatGPT is operating across the world, cultural sensitivities to different regions are crucial to avoid biases (Dwivedi et al., 2023). When AI is used to assist in decision-making across different stages of employment, biases and opacity may exist (Chan, 2022). Stereotypes about specific genders, sexual orientations, races, or occupations are common in recommendations offered by generative AI. Hence, the representativeness, completeness, and diversity of the training data are essential to ensure fairness and avoid biases (Gonzalez, 2023). The use of synthetic data for training can increase the diversity of the dataset and address issues with sample-selection biases in the dataset (owing to class imbalances) (Chen et al., 2021). Generative AI applications should be tested and evaluated by a diverse group of users and subject experts. Additionally, increasing the transparency and explainability of generative AI can help in identifying and detecting biases so appropriate corrective measures can be taken."