Proactive Systems in HCI and AI: Concepts, Challenges, and Opportunities

Technical Summary: Proactive Systems in HCI and AI

Problem Statement
Despite a surge in interest driven by advances in AI, natural language processing (NLP), and sensing technologies, the concept of "proactive systems" remains loosely defined and inconsistently applied across research and practice. Current usage frequently conflates fundamentally different system behaviors; for instance, simple reminders or recommendation systems are often labeled as proactive, even though their underlying mechanisms and intentions differ significantly from systems that genuinely anticipate user goals and act without explicit input. This conceptual ambiguity hampers the ability to systematically design, compare, and evaluate such systems.

Furthermore, existing design and evaluation methodologies are largely rooted in reactive interaction paradigms. Standard tools like the System Usability Scale (SUS) and the User Experience Questionnaire (UEQ), validated primarily in reactive contexts, fail to capture critical nuances of proactive behavior. These include the timing and appropriateness of unsolicited actions, the perceived alignment of system actions with user goals, user control and consent, and the specific dynamics of trust in autonomous decision-making. Current Human-Computer Interaction (HCI) guidelines only marginally address proactivity, leaving a critical gap in standardized approaches for designing and evaluating proactive AI systems.

Methodology
This paper outlines the proposal for a multidisciplinary workshop designed to address these conceptual and methodological gaps. The methodology is structured as a collaborative, interactive event rather than a traditional experimental study. The workshop brings together researchers and practitioners from HCI, AI, and related fields to engage in the following activities:

• Conceptual Framing: Small group discussions to distinguish proactivity from related concepts such as autonomy, automation, and system-initiated interaction, aiming to develop shared working definitions.
• Knowledge Synthesis: Presentations of prior research followed by guided plenary discussions to compare perspectives and identify critical dimensions for study.
• Scenario-Based Design: Participants will create and analyze scenarios involving current and speculative future contexts of proactive systems.
• Evaluation Strategy Development: Groups will exchange scenarios to identify key design considerations (e.g., timing, transparency) and propose potential evaluation approaches or metrics (e.g., trust, user comfort).

The workshop is organized by leading researchers from the University of Toronto, University of Waterloo, and Carleton University, leveraging their collective experience in organizing workshops at major conferences (CHI, HRI, IUI, CUI).

Key Contributions
The paper does not present empirical results or a finalized framework, as the workshop is a proposed event. Instead, its primary contributions are the identification of specific gaps in the field and the establishment of a roadmap for future community building. The workshop aims to:

1. Develop a Shared Conceptualization: Define what constitutes proactive behavior and how it differs from autonomy and automation.
2. Map the Design Space: Identify key dimensions, challenges, and contextual factors that shape how proactive behavior is designed, experienced, and perceived.
3. Establish Guidelines: Co-create human-centered principles and evaluation strategies that account for unique proactive challenges, including timing, transparency, user control, and trust.

Results
As this is a proposal for a future event (scheduled for GI '26 in June 2026), the paper does not contain experimental results. The anticipated outcomes, as described in the "Post-Workshop Plans," include:

• Documentation of key discussion points and insights on an open online platform.
• The identification of concrete research questions and potential collaborations.
• The creation of a dedicated Discord channel to sustain community engagement and facilitate cross-disciplinary learning.

Significance
The paper argues that the significance of this work lies in advancing robust and consistent frameworks for designing and evaluating proactive technologies. By moving beyond superficial manifestations of system initiation (such as suggestions or prompts) to focus on the underlying capabilities that enable effective proactive behavior, the workshop seeks to clarify the conceptual landscape of proactivity. The authors posit that establishing rigorous definitions and evaluation metrics is a prerequisite for realizing the potential of proactive agents to increase perceived helpfulness, improve task efficiency, and calibrate trust in both personal and professional contexts. The initiative aims to bridge the disconnect between the rapid advancement of AI capabilities and the lagging development of human-centric design and evaluation standards.