December 3-5, 2025: GenAI in Localization

GenAI in Localization is a specialized event that is designed for language and localization professionals who want to understand AI adoption, build technical and communication skills, and lead the change.

Powerling and CrossLang collaborated to give a talk about how bias in LLMs can be detected and mitigated. As large language models (LLMs) continue to play an increasingly prominent role in society, growing attention is being paid to the ethical issues surrounding their use, particularly the problem of bias. While significant progress has been made in understanding and addressing these issues, challenges remain in how to detect, measure, and effectively mitigate bias in practice.  These models often reflect and, in many cases, amplify societal biases present in the data used to train them. In this interactive, hands-on session, participants had the opportunity to empirically examine how bias manifests in LLMs, explore techniques for detecting and quantifying bias, and test practical strategies for bias mitigation. The session helped participants critically evaluate bias in LLMs and promote fairer, more inclusive language models.