Description

Amazon Research Awards (ARA) supports academic and nonprofit research that assesses and improves AI across underrepresented languages and cultural contexts. The call advances Amazon’s New Delhi Frontier AI Impact Commitments by expanding datasets, benchmarks, expertise, and evaluation infrastructure, with particular emphasis on the Global South. Responsible evaluation of safety and fairness, with human judgment in the loop, is central to the program.

Priority areas are low-resource language benchmarks, culturally contextual evaluation, and AI deployment in emerging economies. Projects may address translation, question answering, summarization, conversational AI, local norms, community feedback, Indigenous knowledge, or scalable testing infrastructure. Related interests include synthetic data and augmentation, cross-lingual transfer evaluation, human-in-the-loop methods, multilingual transparency, and evaluation literacy and capacity building.