This post is part of the ongoing series of reading reflections on HCI papers and articles. You can find the other posts ‘here.’
Principles of Mixed-Initiative User Interfaces - of Eric Horvitz CHI 1999
Digital Object Identifier (DOI): https://doi.org/10.1145/302979.303030
Summary
This article discusses various principles for designing Mixed-Initiative interactions and the considerations that have to be made for a successful interaction, handling failure, and continuously improving the interaction by learning the user’s preferences.
Prompts
Why is this article important?
This article presents a set of principles for designing mixed-initiative user interfaces that address systematic problems with the use of agents that may often have to guess about a user’s needs. The authors discuss the consideration of uncertainty and the expected costs and benefits of taking autonomous action in different situations. They highlight methods and design principles with examples drawn from the LookOut system.
The authors have listed what they believe are the fundamental principles of a competent mixed-initiative user interface and have given examples of how these principles are implemented in the LookOut system. Some principles discussed are: Scoping precision to match uncertainty, Avoiding the “all or nothing” trap, and Providing a “safety net” for users. The authors also discuss the importance of timing in mixed-initiative user interfaces and how the LookOut system uses timing to determine when to provide services to users.
What do the authors seem to be assuming about the future of A.I. and Human Interaction?
The authors seem to be assuming that A.I. will be able to provide services to users in a way that is more natural and less intrusive than current A.I. systems. They also are hopeful about the potential of learning from users to personalize A.I. systems to the individual user to accommodate their needs better.
I think that mixed-initiative systems have become ubiquitous in the domain of emails and scheduling, with almost all mail and calendar providers having a competent implementation of the same in their email clients (e.g., Gmail, Outlook)
How might you integrate the reading into your academic/professional work?
I think the principles discussed in this paper are fundamental when designing mixed-initiative systems, as they can be very intrusive and annoying to users if not appropriately designed. I think the authors have done an excellent job identifying the principles that should be followed when designing such systems.
The principles I found the most interesting were: Scoped precision to match uncertainty, allowing efficient invocation and termination of services, and the system having sensible and domain-appropriate defaults.
I will use these principles to guide my design decisions when designing user interaction systems.