Assistant Professor Jiaqi Ma has received a grant from the Amazon-Illinois Center on AI for Interactive Conversational Experiences. AICE is a collaboration between Amazon and the University of Illinois Urbana-Champaign to develop intelligent conversational systems that demonstrate contextual understanding and emotional intelligence, allow for personalization, and can interpret non-verbal communication while being ethical and fair.
Ma's research broadly focuses on the data foundations of artificial intelligence, including understanding how training data influence AI models, developing data-centric methods to improve the quality and safety of training data, and studying the broader societal implications of data in AI systems. His AICE project, "A Data Regularization Framework for Data- and Compute-Efficient LLM Post-Training," is funded with a $90,000 award.
The project addresses a fundamental challenge in training large language models: high-quality data are often limited and expensive to obtain, while much larger datasets may be cheaper to collect but contain noisy or inaccurate information. Rather than choosing between the two, Ma's research explores how a small amount of high-quality data can be used to guide and regularize learning from a much larger dataset, improving both data and compute efficiency.
Ma presented his research at AICE's 2026 fall symposium in September.