Rethinking AI Through a Human Lens

At COMET Lab, we spent the past year asking a big question: How do we make AI that truly works for people? Not just functionally, but ethically, transparently, and meaningfully. Our team of 9 came together from diverse backgrounds, united by the belief that AI should be human-aware. Through months of deep exploration into how people experience and interact with AI, we developed a set of clear, actionable guidelines that put users at the heart of AI design.

Limitations of the Double Diamond Framework for AI

Traditionally, UX design leverages the Double Diamond framework—Discover, Define, Develop, Deliver—to create predictable and controlled user experiences.

However, AI design presents unique challenges that this framework struggles to address. AI systems continuously learn and adapt, often yielding unpredictable outputs. For instance, a voice assistant might interpret commands accurately today but might deliver inconsistent results tomorrow as it learns from ongoing interactions. Traditional design cycles that emphasize a linear process fall short when it comes to accommodating this dynamic behavior, requiring a shift toward iterative, adaptive processes with continuous user engagement and feedback loops.

Research: Building Foundations

Our process began with rigorous research:

• Academic Literature Review: We examined over 12 scholarly papers on human-centered AI, extracting themes like transparency, trust, autonomy, and adaptability.

• Expert Interviews: We conducted insightful discussions with professors from renowned universities, exploring cutting-edge practices and challenges in AI design.

• Online Resources: Extensive exploration of industry standards, case studies, and practical examples provided additional real-world context.

Our extensive research phase helped us pinpoint critical areas needing attention in AI design, emphasizing trust-building, clarity, ethical considerations, and adaptive design.

Seminar: Ideating Together

To enrich our findings, we hosted an interactive seminar that invited community insights and facilitated collaborative brainstorming. Participants shared their diverse perspectives, which significantly contributed to shaping our understanding of practical AI challenges and opportunities.

Final Themes and Guidelines

Synthesizing our research, interviews, and seminar insights, we distilled our findings into actionable themes:

These guidelines aim not just to inform COMET Lab's future projects but also to contribute meaningfully to the broader landscape of ethical and human-centered AI design.

Our work reinforces a critical shift—designing for AI is about embracing uncertainty, fostering transparency, and empowering users in every interaction.