How design adds value when AI is the new UI (AI summit Amsterdam)
Author
Willem Duijvelshoff
Published
16 October 2018
Reading time
3 minutes
Aside from being one of the biggest conferences in the world on artificial intelligence (AI), the AI Summit in Amsterrdam differentiates itself by not just focusing on technological possibilities. During the two-day event, a lot of attention went to the societal and ethical challenges around AI innovation. This resulted in a unique collection of talks and workshops, stimulating thought about the future of product development and design.
AI through the eyes of design
Currently, AI is becoming increasingly accurate in solving narrow tasks like image recognition and language processing. It’s becoming common ground that APIs are offering intelligence as ‘plug and play’ services through the cloud. Attending speakers from Microsoft and Google offer these services and tried to persuade developers to start using their APIs.
Technology stacks aside, Mark Wiermans (Accenture) opened the event by stating ‘AI is the new UI’, a statement that made me ponder the added value of design when user interfaces take on different forms. This sparked the central question I aimed to answer during the conference: What role do designers play in the age of AI? And how can our human-centered way of thinking and acting add the most value?
Answering the right question
AI is at the center of the hype at the moment. Companies are bringing data scientists and AI experts in-house. But blindly following the hype brings the risk that teams start analysing data, but without a plan they will generate limited value. Bram Kokke (AkzoNobel) summarized the issue and presented a solution: Look at the pain points of customers instead of the hype or technology. This ties into design thinking, starting with customers and their needs is at the core of the design profession. In this case the challenge for designers is to help data teams to start answering the right questions.
Luciano Floridi (Oxford Digital Ethics Lab): The key is to realize a preferable product
This can be done by mapping the user journey and related tasks and seeing where AI can contribute. Santiago Braje and Androniki Menelaou (ING) worked on Katana, an investing tool that is designed to help decision-making by traders without replacing them. With design thinking, they mapped the decision-making processes of the traders to uncover where AI can add the most value, generating insights that made their AI solution more customer centric and potentially more successful in the long run.
Humans first, AI second
Responsible and ethical AI was an often recurring theme in workshops and talks. It occurred to me that ethics can be one of the areas in which the design methodology can be of great value as well. During the workshop “Shaping Values for our Future with AI” by the United Nations Secretary Panel on Digital Cooperation, groups of attendees quickly brainstormed their hopes and fears concerning AI. The moderators helped them to cluster these hopes and fears so that common values could emerge (resulting in values like safety and self-actualisation).
In the early stage of product development, these co-creation methods can be used to formulate design principles, designers, researchers and developers can refer to these values when facing ethical dilemmas. In a later talk, Ajun Moon referred to this as an ‘ethical road map’. If you formulate the core values you want to underpin your AI product, this road map can help you make ethical decisions in later stages of the development process.
Prototyping ethical probes
Assessing the user’s ethical framework and how they perceive a proposed AI technology can be done with ethical probing. An ethics probe interrogates the public attitude, the team of Niels Wouters (University of Melbourne) prototyped a biometric mirror, that uses facial recognition software to analyze the facial features of the user while determining their ‘psycho-metric characteristics’ based on their face. This analysis of course wasn’t accurately describing their character, but the prototype was meant to provoke and seek out a response.
The values show supposed attributes that are uncovered with facial recognition
This shows how prototyping and experimenting with AI services can feed into the design process, these kind of prototypes can map out the sentiment towards upcoming technologies assessing their feasibility. It can also lead to what Luciano Floridi calls soft ethics, ethics that help shaping products in positive ways. Not by saying “you shouldn’t do A or B” but by uncovering the users ethical framework and using this as a constructive input for the design process.
Concluding
My underlying sentiment was that a human approach to AI innovation will still need design. Interfaces and touchpoints might change, but maximizing customer value and the benefits of a human-centered process remain. AI brings about a lot of fears of replacement of jobs and human tasks being automated, but AI shouldn’t be about replacing humans but about amplifying their capabilities. Let’s start designing this future together.