Our 21st century world is driven by growing complexity, data volume and information. And AI is the epicentre of this evolution. Smart machines are increasingly intertwined with our lives as we count on them to process repetitive tasks and make routine decisions on our behalf. And the power of AI is growing. Today, it powers 25% of customer service operations and McKinsey estimates it will drive around US$13 trillion in new global economic activity by 2030.
New Directions
AI is shaping a new reality. In this process, design is having to bridge the gap between it and science. And become more multidisciplinary and agile to stay at the top of its game as a creative solutions provider. As designers, we’re moving out of a tactical mindset to one more strategic, where there are strong human and creative elements in the picture. And, as AI becomes more embedded in daily life, the ethical dimension rises to the top, where designers must imbue machine systems with empathy and judgment free of bias.
How should we design AI and AI experiences with voice, visuals, gestures and complex invisible interfaces? And, as an industry, how should designers collaborate with data scientists and engineers strike the right balance to advance the future for informed AI systems?
In this, our third virtual DesignMeets session, we look at the current AI landscape and the changing role of design in its growth and development. How will design respond to the acceleration of tech and the explosion of AI and what is its impact on us and the design community?
The Panel
Our first speaker is Chloe Benaroya, an innovation and human-centred design coach who’s worked in instructional design and led multidisciplinary teams on digital platform development teams, educational games and apps over the past 20 years. She’s currently consulting at Manulife Financial’s Innovation and UX Lab in Montreal.
“I prefer using words like ‘smart or intelligence augmentation’,” she says, as they more accurately reflect the kind of projects she typically works on, involving people. To demonstrate, she profiles a typical AI project involving designers that uses natural language processing (NLP) in voice-activated apps such as chatbots and smart/adaptive FAQs, to convey meaning from human language for decisions. And NLP tools can design interactions without coding or software.
In these projects, designers build value (creating benefits to the listener) and help build the knowledge database over time, which is how the information is gathered, stored and modified to stay relevant. The next step in the process is understanding context (asking the right questions to get the expected outcomes) and defining interactions. For designers, this involves critical exploratory user research and working within the NLP tool limitations (size of memory etc.) and parameters requiring close collaboration with technology partners and the product development team. And last, testing interactions by pushing the boundaries of the new technology, monitoring the NLP system in action and iterating for improvement.
Fanny Sie (FS) is Head of Artificial Intelligence and Digital Health at Hoffmann-La Roche Canada. As a clinician, researcher and business developer in AI, analytics and digital health assets, her broad experience has contributed to new products and services coming onstream in the growing healthcare market.
Roche’s mandate is to make technology a progressively positive impact in people’s lives. Largely due to COVID-19, the company has been transformed from being pharma-centric to developing people and health solutions, dissolving siloes of different disciplines (scientific, healthcare professionals and product & service delivery), focusing on user experience and the patient journey. Everything from waking up in the morning, their daily activities and what they do to manage their healthcare.
As part of this initiative, the Roche Data Science Coalition was formed under the Roche umbrella, bringing together public and private groups to make the world a better place with open data available to create solutions for general public healthcare worldwide. To expand the benefits of this model in more therapeutic areas, the Roche AI Centre of Excellence (AI with Roche or “aiR”) was created specifically to focus on leading healthcare challenges and problem-solving in Canada and around the globe.
“From the design perspective, we’re creating a way for people to interact with one another and accelerate discovery,” she says, “while at the same time translating it into products and services, market policies, protocols and technologies that all work for patients faster.”
Our final panellist is Miki Arai (MA), who leads the design practice focusing on strategic information design, data visualization and UX design at Omnia AI, Deloitte Canada’s Artificial Intelligence studio. Bringing a user-centred design lens to complex challenges intersecting business, data and technology, Miki’s impactful design projects have ranged from digital applications and web design to brand communication collateral design for clients in a cross-section of industries.
AI plays an increasing part of people’s lives but it’s still a mystery how data is collected, used and how it actually works. Design focuses on the human component, focusing on user-centred design thinking, information design and data visualization in the role of the designer in building AI solutions.
Its primary role is that of an interpreter, sorting and arranging data from the human point of view, making it visually productive and telling a story. Designers translate complex data models into relatable, tangible and accessible output that bridges the gap between data output and human consumption. They help define the purpose for AI, in concert with data scientists and engineers, to create the human dimension for AI. Design is the empathetic part of the puzzle (desirability) for user needs, distinct from the economic (feasibility) and technical (viability) perspectives that must all come together seamlessly to arrive at the sweet spot of innovation.
“Our brain is visual,” she says, “so it’s super important to pay close attention to the visual components of output.” With good UI design, visual basic information is easier to process, actionable insights are more accessible and this leads, cognitively speaking, to better decisions.
Selected Q&A
What’s your journey been like from graphic design to analytics, UX and AI?
— MA: My journey started off in traditional graphic design, working at a studio with a lot of academics and having a lot of information thrown at me. That led to understanding information and presenting this information to a non-academic audience, which took me to analytics practice with not much of a data perspective but a natural curiosity and talking to different people in this space. Analytics is the first step toward AI. You start with a good understanding of graphic design principles, the ability to deliver good design and being able to understand the problem. This brought me to human-centred design, design thinking and, ultimately UX. And, of course, natural curiosity and understanding of others. Always think of what you can bring as a person and contribute.
Increasingly, technology companies are looking at the ethical dimensions of AI and other data-driven products. Are any such initiatives happening in your own organizations?
— FS: Ethical considerations are most important, particularly in healthcare. They are our number one priority as we’re dealing with people and the software and algorithms we’re using may potentially assist in a person’s diagnosis and care plan. We take these issues very seriously throughout our entire development all the way from concept through to prototype, scaling and implementation, particularly in the areas of privacy and security.
From your experience, how do you create ethical and non-discriminatory AI that’s being fed through human-gathered data with biases and limitations?
— FS: The answer for us lies in transparency. If we’re going to use a particular algorithm in standard practice, particularly in medicine, we want to know the conditions in which it was made. We want to see the model, the data points and we should have access to another data set to further validate the information. So, the overall answer is transparency, no black boxes and being cognizant of biases is important, and how bias actually reflects the types of algorithms and their potential use in the environment.
The Takeaway
Since the very early days of AI, when the term was first coined by Alan Turing in the 1950s, artificial intelligence has grown exponentially as an influence on the nature of our society. Today, the growing impact of data, technology and the explosion of AI as formative influences in our lives is equally profound.
As designers, we bear an important responsibility for the future of machine learning and ensuring that its many perspectives (economic, ecological, sociological) are empathetic and user-focused, infused with creative thinking, free of discrimination and bias. It is our opportunity to play a pivotal role in this process by navigating the coming challenges and instilling the human dimension within these new smart systems.
Sign up for our newsletter to stay in the loop.