Cross-cultural challenges in generative AI: Addressing homophobia in diverse sociocultural contexts
Bibliographic record
Abstract
Previous discussions have highlighted the need for generative AI tools to become more culturally sensitive, yet often neglect the complexities of handling content about marginalized groups, who are perceived differently across cultures and religions. Our study examined the responses of two generative AI systems to homophobic statements and explored how their outputs varied when different societal and religious context information about the user was provided. Findings showed that ChatGPT 3.5's replies frequently reflected cultural relativism, as evidenced by an emphasis in the outputs on the idea that different cultures hold distinct perspectives and that these diverse viewpoints should be respected. In contrast, Bard's responses often stressed human rights and provided more support for gay people and lesbian, gay, bisexual, trans, and queer (LGBTQ)+ issues. Both systems demonstrated significant variation in their responses depending on the contextual information provided in the prompts, suggesting that AI systems may adjust the degree and form of support they express for LGBTQ+ people and issues according to the information they receive about a user's background. While our analysis focused specifically on chatbot responses to homophobic statements, the study underscores a broader dilemma concerning the tension between cultural relativism and universal human rights in generative AI—an issue that extends beyond homophobia to include animosity toward other marginalized groups that are perceived differently across societies and religions. The study contributes to understanding the social and ethical implications of AI responses and argues that any work to make generative AI outputs more culturally diverse requires grounding in fundamental human rights.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.031 | 0.045 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.007 | 0.010 |
| Scholarly communication | 0.008 | 0.006 |
| Open science | 0.002 | 0.012 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".