The field of human-large language model (LLM) collaboration is rapidly evolving, with a focus on developing more intuitive and effective interfaces for users. Recent research has highlighted the potential of LLMs to revolutionize virtual experiences, improve user engagement, and facilitate creativity. A key direction in this field is the development of proactive information gathering capabilities, which enable LLMs to identify gaps in context and elicit implicit user knowledge through targeted questions. Another important area of research is the creation of user-centric benchmarks and environments, such as interactive gym environments, to evaluate and advance the ability of LLMs to collaborate with users. Noteworthy papers in this area include: Talking-to-Build, which explores the impact of LLM-assisted interfaces on player performance and experience in Minecraft. Teaching Language Models To Gather Information Proactively, which introduces a new task paradigm for proactive information gathering and demonstrates significant improvements in LLM performance. UserBench, which provides an interactive environment to measure and advance the capability of LLMs to collaborate with users. IntentFlow, which supports the communication of dynamically evolving intents throughout LLM-assisted writing. Mitigating Response Delays in Free-Form Conversations with LLM-powered Intelligent Virtual Agents, which investigates the challenges of mitigating response delays in free-form conversations with virtual agents powered by LLMs. User Feedback in Human-LLM Dialogues, which studies harvesting user feedback from user-LM interaction logs and provides insights into the potential and limitations of implicit user feedback.