Banks must address bias in large language models

All cultures have explicit and implicit biases. Preventing discrimination when large language models are involved starts with using high-quality human-curated data and text.