top of page

User-Centered AI
Design Playbook

Role: Lead UXR, Principal Designer, Data Analyst 

As AI rapidly advances in everyday life, it's important to keep the human user at the center of design. AI interactions can be complex, and features like automation raise important concerns around trust, privacy, and ethics.

The objective was to develop a comprehensive UX playbook that distills best practices, design patterns, and heuristics for building user-centered AI features—ensuring transparency, accountability, usability, and ethical alignment across product teams.

Schlumberger_2022.webp

The Problem

  • Product teams were beginning to integrate generative AI into their workflows without consistent standards, leading to fragmented approaches to usability, ethics, and trust.

  • Without clear guidelines, AI features risked confusing users, producing biased or misleading outputs, and eroding confidence in the technology.

  • Existing UX heuristics did not fully address the unique challenges of AI, such as explainability, automation, and personalization.

Goals

  • Create a comprehensive, user-centered AI design playbook that translates abstract principles into actionable tools and guidelines.

  • Provide practical resources—checklists, heuristics, and do/don’t examples—that teams can apply throughout the product lifecycle.

  • Benchmark leading AI tools to identify best practices and gaps, grounding the playbook in real-world evidence.

  • Establish a shared framework that improves consistency, reduces risk, and ensures AI features are transparent, ethical, and user-friendly.

Research Questions

  • What core UX principles are most critical for designing AI systems that are transparent, trustworthy, and accessible?

  • How can these principles be translated into practical guidelines, checklists, and do/don’t examples that product teams can apply at different stages of development?

  • What do leading generative AI tools (e.g., ChatGPT, Gemini, Copilot) already do well, and where are there gaps or inconsistencies that inform design opportunities?

  • How can heuristic evaluation methods be adapted to identify usability, fairness, and ethical risks specific to AI interfaces?

Methods

image.png

  • Conducted secondary research drawing from academic literature, industry standards, and public documentation from leading AI companies.

  • Performed a thematic analysis to group findings into eight core design principles: transparency, accountability, usability, personalization, human–AI interaction, fairness, security, and sustainability.

  • Created heuristics and evaluation worksheets to help teams identify usability and ethical risks in AI interfaces.

  • Compiled a set of do’s/don’ts and quick-start checklists to translate principles into actionable design guidance.

  • Completed a comparative analysis of major AI tools (ChatGPT, Gemini, Copilot, etc.) to benchmark best practices and gaps in real-world products.

Results

image.png

  • Delivered a UX AI Design Playbook adopted as a practical reference for responsible AI design.

  • Produced eight actionable design principles with accompanying heuristics, helping teams evaluate and improve AI features consistently.

  • Developed phased checklists (early, middle, late) tailored to product maturity, ensuring teams can apply the guidance at any stage.

  • Highlighted strengths and weaknesses across industry-leading tools, enabling teams to learn from benchmarks and avoid repeating common pitfalls.

  • Established a shared framework that improves user trust, reduces design risk, and promotes ethical, user-centered AI development.

Impact

The playbook provides practical, user-centered guidelines for teams developing AI products, serving as both a checklist and reference throughout the product lifecycle. By focusing on transparency, ethics, and accountability, it reduces risk, builds trust, and enhances user experience. The framework promotes consistency across teams while also benchmarking against leading AI tools, ensuring design decisions are both grounded in best practices and adaptable to evolving standards.

"This has far exceeded my expectations. I could publish this right now"

SLB Researcher

bottom of page