4 articles tagged with "bias"
Recent research highlights the sensitivity of language models to contextual information, which can lead to harmful biases in decision-making. Direct preference optimization offers a potential solution.
An international study indicates that police violence is perceived as less severe when the victim has an immigrant background, highlighting systemic biases in societal views.
A study led by Aliah Zewail reveals that large language models exhibit predictable stereotypes related to non-Western moral values, raising concerns about AI ethics.
This analysis delves into the implications of AI biases on media content, emphasizing the need for transparency in algorithmic processes.