Design for uncertainty; an approach to designing real-time experiences for AI and ML systems

As Artificial Intelligence and Machine Learning (AI/ML) continue to advance, designers are faced with the challenge of shaping user experiences that foster understanding and trust in these transformative systems. AI/ML systems, driven by data and predictions, inevitably encounter imperfections and uncertainties within their training data, leading to potential errors. However, it is the responsibility of designers to craft experiences that empower users to comprehend the decisions made by AI/ML systems and instil confidence in their outputs.

In this presentation, I will draw upon real-life experiences, specifically focusing on the design of speaker diarization, which is the first real-time speaker identification feature designed for Deaf and Hard of Hearing individuals. I will share insights aimed at prioritizing user understanding and trust in AI systems. By placing the user's experience at the forefront, we can design interfaces that are not only intuitive but also adaptable to ever-changing circumstances. This approach ensures that users can confidently interact with these complex systems. Whether you are a designer, developer, or simply intrigued by the future of technology, this talk will provide you with indispensable principles that will guide the creation of dynamic experiences for AI and ML systems in real-time. These experiences will perform reliably and effectively, even in the face of unpredictable or uncertain conditions.

Let us embark on this journey together, shaping a future where intuitive design and trust in AI/ML systems intertwine harmoniously.

As an experienced product designer with a passion for Artificial Intelligence (AI), Machine Learning (ML), and inclusive design, I have developed a unique approach to design that draws from my background in designing for AI and people with special needs.

Previously, as a designer at Ava, one of my notable achievements involved designing speaker diarization, the first speaker identification feature for Deaf and Hard of Hearing people. This groundbreaking technology uses AI to identify speakers in real-time during conversations, making it easy for Deaf and Hard of Hearing people to follow along.

I recently spoke about this topic at the just concluded Open Source Community Africa (OSCA) held in Lagos, Nigeria, last month and I have an already published article on Medium discussing the design principles and practical approaches that can help designers create highly dynamic experiences for AI/ML systems that can perform reliably and effectively in uncertain conditions.

Technical requirements for participants include; a thorough understanding of data science and AI terminologies and techniques as we'll be discussing the ethical standards and pitfalls of building accountable, fair, transparent, and trustworthy AI.