The paper "Designing for Human-Agent Alignment: Understanding what humans want from their agents" by Nitesh Goyal, Minsuk Chang, and Michael Terry explores the alignment between human users and autonomous agents, particularly in the context of selling a used camera online. The authors conducted a qualitative study with 10 participants to understand the parameters needed for effective human-agent collaboration. The study identified six key dimensions of alignment: 1. **Knowledge Schema Alignment**: Agents should gather and provide relevant information early in the negotiation process. 2. **Autonomy and Agency Alignment**: Agents need clear boundaries and guidelines for when to act autonomously and when to seek human input. 3. **Operational Alignment and Training**: Agents require training and guidance on negotiation strategies and ethical behavior. 4. **Reputational Heuristics Alignment**: Agents should consider the impact of their actions on the user's reputation. 5. **Ethics Alignment**: Agents must adhere to ethical standards and avoid behaviors that could harm the user. 6. **Human Engagement Alignment**: Agents should engage with users at appropriate times and provide meaningful interactions. The findings highlight the complexity of aligning human and agent behaviors and the need for designers to consider these dimensions in the design of autonomous agents. The authors also discuss three design directions for improving human-agent collaboration, emphasizing the importance of human-centered research and the need for agents to self-reflect and learn from past interactions. The study underscores the importance of aligning agent behavior with user values and safety, and suggests that agents can play a role in ensuring appropriate alignment through simulations and feedback mechanisms.The paper "Designing for Human-Agent Alignment: Understanding what humans want from their agents" by Nitesh Goyal, Minsuk Chang, and Michael Terry explores the alignment between human users and autonomous agents, particularly in the context of selling a used camera online. The authors conducted a qualitative study with 10 participants to understand the parameters needed for effective human-agent collaboration. The study identified six key dimensions of alignment: 1. **Knowledge Schema Alignment**: Agents should gather and provide relevant information early in the negotiation process. 2. **Autonomy and Agency Alignment**: Agents need clear boundaries and guidelines for when to act autonomously and when to seek human input. 3. **Operational Alignment and Training**: Agents require training and guidance on negotiation strategies and ethical behavior. 4. **Reputational Heuristics Alignment**: Agents should consider the impact of their actions on the user's reputation. 5. **Ethics Alignment**: Agents must adhere to ethical standards and avoid behaviors that could harm the user. 6. **Human Engagement Alignment**: Agents should engage with users at appropriate times and provide meaningful interactions. The findings highlight the complexity of aligning human and agent behaviors and the need for designers to consider these dimensions in the design of autonomous agents. The authors also discuss three design directions for improving human-agent collaboration, emphasizing the importance of human-centered research and the need for agents to self-reflect and learn from past interactions. The study underscores the importance of aligning agent behavior with user values and safety, and suggests that agents can play a role in ensuring appropriate alignment through simulations and feedback mechanisms.