A new study demonstrates that advocacy groups can influence the values and outputs of large language models by making strategic edits to Wikipedia articles. Researchers found that Wikipedia's high weighting in LLM training data means that even small, coordinated editing campaigns can measurably shape how models discuss specific topics, such as animal welfare. This approach offers a low-cost method for organizations to influence AI systems' perspectives. AI
IMPACT Advocacy groups can leverage Wikipedia editing to shape AI model outputs on specific topics, offering a new avenue for influence.
RANK_REASON The cluster reports on a study detailing a new method for influencing AI model behavior, which falls under research. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
