University Of Massachusetts Amherst Assistant Professor Receives Microsoft AI And The New Future Of Work Award

Hamed Zamani, assistant professor in the Manning College of Information and Computer Sciences (CICS), has been selected to receive a Microsoft AI and the New Future of Work award in support of his project, “Improving Productivity by Personalizing Large Language Models,” which aims to introduce personalization to large language models (LLMs) to enhance user productivity.

LLMs have significantly influenced the reshaping of Natural Language Processing (NLP), playing a key role in user-facing applications that improve productivity, including question-answering systems, writing assistants and task management tools. However, the need for personalization that addresses users’ unique needs and behaviors remains largely unexplored.

Recognizing that user needs and behaviors vary widely, the team of researchers in the UMass Amherst Center for Intelligent Information Retrieval (CIIR), led by Zamani, will utilize its previously developed Language Models Personalization (LaMP) benchmark, a comprehensive measure designed to assess the personalization capabilities of language models, to investigate how to personalize LLMs to meet individual requirements more effectively, while increasing user satisfaction and productivity. The team will also employ a retrieval-augmentation solution that retrieves personalized items from a user’s profile and incorporates them into the LLM prompt to create a more tailored experience while maintaining privacy standards.

Building on their work applying the LaMP benchmark, the team will explore the impact of LLM personalization on productivity and user satisfaction in real-world tasks, such as email writing and in-context question answering. They will also develop time-aware retrieval models which take into consideration the recency of user activity for even more effective personalization.

Microsoft’s AI and the New Future of Work award aims to support research that explores how large language models and AI systems can maximize human productivity, including the use of LLMs in collaborative work, technical and non-creative writing, chatbot engagement and workplace communication. Nine proposals, selected from a pool of over 150 worldwide, received funding in 2023.

Zamani, who joined the CICS faculty as an assistant professor in 2020, is associate director of the Center for Intelligent Information Retrieval. His current research focuses on information retrieval, conversational search and retrieval-enhanced AI. He received his doctorate in computer science in 2019 from CICS and his master’s and bachelor’s degrees in computer engineering from the University of Tehran in 2015 and 2013 respectively.