The paper "Principles of Mixed-Initiative User Interfaces" by Eric Horvitz from Microsoft Research explores the integration of automated services with direct manipulation in user interfaces. The author reviews principles that can enhance human-computer interaction by combining these two approaches effectively. Key challenges and opportunities in designing mixed-initiative interfaces are discussed, emphasizing the need to avoid focusing solely on either automated services or direct manipulation. The paper highlights the LookOut system, which overlays automated scheduling services on Microsoft Outlook. LookOut identifies new messages and attempts to assist users with calendar management and appointment scheduling. It uses probabilistic models to infer user intentions and decides whether to engage in a dialog or take automated actions based on the expected utility of different outcomes. The principles for mixed-initiative UI include: 1. Developing significant value-added automation. 2. Considering uncertainty about user goals. 3. Considering the status of user attention in the timing of services. 4. Inferring ideal actions based on costs, benefits, and uncertainties. 5. Employing dialog to resolve key uncertainties. 6. Allowing efficient direct invocation and termination. 7. Minimizing the cost of poor guesses about action and timing. 8. Scoping precision of service to match uncertainty and variation in goals. 9. Providing mechanisms for efficient agent-user collaboration. 10. Employing socially appropriate behaviors. 11. Maintaining working memory of recent interactions. 12. Continuing to learn by observing. The paper also discusses decision-making under uncertainty, using decision theory to determine when to take action versus inaction, and the role of user attention in the timing of services. LookOut's ability to handle invocation failures and its use of probabilistic models to infer user goals are also detailed. The paper concludes by emphasizing the potential for fundamental enhancements in human-computer interaction through the integration of automated services and direct manipulation.The paper "Principles of Mixed-Initiative User Interfaces" by Eric Horvitz from Microsoft Research explores the integration of automated services with direct manipulation in user interfaces. The author reviews principles that can enhance human-computer interaction by combining these two approaches effectively. Key challenges and opportunities in designing mixed-initiative interfaces are discussed, emphasizing the need to avoid focusing solely on either automated services or direct manipulation. The paper highlights the LookOut system, which overlays automated scheduling services on Microsoft Outlook. LookOut identifies new messages and attempts to assist users with calendar management and appointment scheduling. It uses probabilistic models to infer user intentions and decides whether to engage in a dialog or take automated actions based on the expected utility of different outcomes. The principles for mixed-initiative UI include: 1. Developing significant value-added automation. 2. Considering uncertainty about user goals. 3. Considering the status of user attention in the timing of services. 4. Inferring ideal actions based on costs, benefits, and uncertainties. 5. Employing dialog to resolve key uncertainties. 6. Allowing efficient direct invocation and termination. 7. Minimizing the cost of poor guesses about action and timing. 8. Scoping precision of service to match uncertainty and variation in goals. 9. Providing mechanisms for efficient agent-user collaboration. 10. Employing socially appropriate behaviors. 11. Maintaining working memory of recent interactions. 12. Continuing to learn by observing. The paper also discusses decision-making under uncertainty, using decision theory to determine when to take action versus inaction, and the role of user attention in the timing of services. LookOut's ability to handle invocation failures and its use of probabilistic models to infer user goals are also detailed. The paper concludes by emphasizing the potential for fundamental enhancements in human-computer interaction through the integration of automated services and direct manipulation.