Google Cloud
Speculative Futures of Human–AI Collaboration in Customer SupportDiscover
Google Cloud faces a critical scaling challenge: a growing customer base with increasingly complex needs that human resources alone cannot meet. This research explores how AI can address these limitations by enabling more proactive, human-centered support experiences. Specifically, we will uncover opportunities to anticipate user needs before problems arise and identify design opportunities that integrate advanced analytics, generative AI, and adaptive tools to deliver personalized service at scale.
RQ3: What is the social implication of these alternative futures?
Approach:
Define
Literature Review:The Core Insight: The future of support is not about automation percentages: it's about creating partnerships where AI and humans each do what they do best, with users genuinely able to shape how that partnership works.
User Research:
Key Findings:
Trust, Control
Transparency, Boundaries & Data
Speed, Proactivity & Escalation
Learning, Memory & Context Awareness
Reliability, Safety & Accountability
Users want agency, and control over when and how AI intervenes. They trust humans more in high-stakes or emotional contexts.
Participants demand clarity about what AI sees, records, and acts on.
Fast resolution matters more than who provides it — but escalation must always be visible and easy.
Memory and contextual understanding are essential for trust and usability.
Users demand trustworthy systems that prevent costly mistakes or misinformation.
People value autonomy and boundaries in support systems. Trust develops when AI waits, asks permission, and hands off seamlessly to humans.
Transparency builds comfort. Users accept proactive AI only if it clearly explains visibility and control over data.
Responsiveness is central. AI is welcomed if it accelerates resolution but must never trap users in loops or hide the human option.
Users equate memory with intelligence and respect. Forgetful AI feels unreliable and unempathetic.
AI support must prioritize error prevention and accountability.
Develop
Based on the research findings, we developed four concepts and evaluated them through usability testing with AI-generated workflows and rapid prototyping.
Four Speculative Scenarios:
Based on the testing results, we chose to further develop the Nurse–Doctor metaphor scenario.
Power Dynamic: Clear hierarchy - AI triages, human is expert authority
Core Behavior:
AI is first responder, gathers information, does initial assessment
Human "specialist" only sees pre-diagnosed, prepared cases
AI prepares "patient file" for human
Flow Additions:
AI assessment phase: Categorizes problem severity/type
AI creates briefing document for human before handoff
Branch: "Simple case" → AI handles completely, no human needed
Branch: "Complex case" → AI prepares detailed context for human expert
Post-interaction: Human validates AI's assessment
Deliver
Research Report: https://drive.google.com/file/d/1R917m_9aRI7cWK0eu4pGkU1BTrkhUKdm/view?usp=sharing
Design:
Branding
Demo Video: