Unlocking Smarter Information Access, For Humans and AI Agents Alike
Information access (IA) refers to retrieving, evaluating, organizing, and digesting information, with broad applications such as search engines, recommender systems, and question-answering systems used daily by millions worldwide.
Natural language processing (NLP) has long played a key role in enhancing the intelligence of IA systems. With the rise of large language models (LLMs) and AI agents, there are unprecedented opportunities to develop interactive agents that improve human IA quality and reduce their effort.
At the same time, as AI agents increasingly automate our tasks, the agents themselves have the need for IA, creating new challenges in enabling AI agents to best leverage IA tools, such as search engines to retrieve relevant information to augment generation.
These new opportunities and challenges call for research in the interdisciplinary area of IA, NLP, LLMs, and interactive AI agents. There are many general open questions to be investigated, such as:
- How can we maximize the benefits of AI agents for human IA?
- How can agents acquire complete user context, process it, and deliver accurate, timely, and personalized information services?
- How can AI agents simulate users and understand their needs?
- How can privacy be protected given the extensive user data required for personalized IA?
- How can AI agents utilize existing IA technologies to inform their own decision-making and enable self-improvement?
With a broader view of IA serving both human users and AI agents, this workshop examines the current state of the art and promising future directions in IA paradigms for the era of AI agents:
- For humans, IA will be enhanced in effectiveness and efficiency through AI agent assistance; and
- For AI agents, they can leverage existing human IA technologies to supplement their limited pre-training knowledge, inform decisions, and autonomously improve their capabilities.