Jef Willemyns


Design in the age of AI

Executive summary

Due to the current rate of innovation and development in the field of artificial intelligence there’s a momentum and opportunity not to be missed if you are a designer, manager or data scientist. This evolution will change the field of design and the role of designers forever in a meaningful way. This blogpost discusses the implications and the impact of the rise of artificial intelligence for the field of design and AI. It will influence the process, the principles and the role of designers in the following ways:

  • A split of the design process into a human-intensive part and an artificial intelligence part.
  • A reinforcement of the current design thinking principles. Artificial intelligence offers unprecedented opportunities to automate learning, which is at the core of developing innovative solutions.
  • A change in the role of design(ers). The human side of design will prevail as an activity of “sensemaking”.

How AI impacts the design principles

Contemporary design practice is centered around three pillars, known as the three principles of design thinking. The first principle is that design is people-centered, meaning that innovation, when driven by design, is inspired by empathy with the user. The second principle states that design is abductive. This means that design has a generative approach to create solutions since abduction often leads to reframing the problem and the questions that inform the design. The last principle is that design is iterative. By using fast testing cycles, these abductions are continuously adapted and improved. Prototypes act as a ‘playground’ for the exploration of ideas, for discussion, conversation and learning.

This raises the question following question: Is artificial intelligence going to change how design is practiced, and does it have an implication on the current design principles from design thinking? Will AI question these existing principles or will they be reinforced?

These principles also dictate the limitations of design. The more people-centered a design is, the more difficult it will be to produce it on a certain scale. The bigger the scale, the harder it is to be really people-centered or iterate and adapt in fast cycles.

If we look at those principles, we can see that they are reinforced by the use of AI. In practice, the design thinking principles are not undermined by using artificial intelligence. Rather, by removing past limitations in scale, scope and learning, it enables to further enact design in its core. It realizes the ultimate form of people-centeredness, with experiences that can be designed for each individual person, and continuously improved based on individual user data. AI may enhance creativity, by expanding the scope of the design space beyond product categories and industries. And finally, it brings iterations and experiments at the core of the operating models and firms.

Impact of AI on the design process

Design is traditionally very human intensive. To design implies to take a number of decisions. A few of them require imagination and creativity. Most decisions, however, require more specific problem-solving skills.

AI revolutionizes this scenario. Real time data can be gathered and fed to AI engines which have problem solving capabilities. These can generate specific solutions for individual users autonomously giving designers the opportunity to release themselves from the burden of detailed development. The paper labels this as AI factories where designers conceive new offerings and create the problem solving loop (see image). The problem solving loop itself takes care of the detailed development of each product. They are autonomous and human-capital-free design systems. They are resilient to variations in volumes and can provide a variety of solutions without the R&D efforts. They have two main capabilities: The first one being the gathering, cleaning and normalizing of data, the second one to solve the customers challenge using a set of rules, theories, programs and algorithms that perform intelligent tasks (1).

In practice

One of the examples of AI empowered design in practice is the case of Netflix. They started using AI in 2010 and at the moment, every interface for every singular user is unique and designed in the moment. This means that AI determines what content fits for the user, in which order it is presented and what images are used to do effective recommendations. Netflix uses different types of machine learning algorithms that each resemble parts of the design process.

  • Supervised learning: These algorithms are used to come as close as possible to an accepted outcome. Netflix uses these for recommendations. Just as human designers immerse themselves in the context of users, the algorithms are trained with significant information on the context and the user.
  • Unsupervised learning: The goal of unsupervised learning algorithms is to discover insights in data with few preconceptions or assumptions. Netflix uses unsupervised learning to create customer segments. Unsupervised learning algorithms mirrors ideation and brainstorming as well as the basic perspectives of design thinking.
  • Reinforcement learning: These algorithms need a starting point and a performance function. The tradeoff is whether to spend more time exploring options or exploiting the algorithm. Netflix uses this to determine visuals for recommendations by iterating between the exploration and the exploitation phase. This strongly resembles the diverging and converging phases of design thinking.

The problem solving capabilities of AI lays the foundation for reframing design practice. This has profound implications in terms of the object and of the process of design. Firstly, the object, what a customer experiences, was previously conceived and developed by designers. The box of a VHS cassette was designed and developed down to the level of details by a designer. A Netflix movie recommendation is not. It is designed in the moment by an algorithm deciding what images, position and text would influence a specific user the most, and thus is different for every single user. The designers do not design the solution anymore, they design the problem solving loops.

Secondly, the design process itself changes and is split into two parts. There’s a human intensive design phase to conceive the solution space and the problem solving loops. And then an AI powered phase to develop the specific solution for a specific user which requires virtually zero cost and time. This new practice is clearly visible in digital experiences like Netflix and AirBnB but is gaining traction in the design of physical products as well. Tesla for example is still bound to the limits of hardware but uses the power of AI in two ways. Firstly, they limit the physical user interactions and replace them with software and screens and secondly, they overload their cars with sensors that collect massive amounts of data. This data is used to train their learning algorithms and reinforce datasets. It also allows them to activate new services remotely and after the product release.

Impact of AI on the role of designers

Artificial intelligence automates learning, which is the core of innovation. It offers unprecedented opportunities to reduce the cost and time of developing new solutions. Designers in practice will need new models that support the designing of problem-solving loops that will develop the solutions. The role of designers is shifting towards understanding what problems make more sense to address, next to designing the learning loops, and then to drive their continuous evolution towards a meaningful direction.

For managers, understanding the new nature of the design practice is fundamental to avoid applying the right design principles to the wrong process. There is also a need to strengthen the understanding of the strategic dimension of design, the definition of a meaningful direction. The human side of design becomes an activity of sense-making, defining which problems make sense to address.

As a designer, I am very interested in the implications of AI on my work, so I researched this in my master thesis. In researching the most average and general shape of a chair, I used artificial intelligence, and more specifically, a generative adversarial network. This is a neural network that learns to recognize chairs in a dataset of images and tries to reproduce them. This dataset creates the search space or the boundaries in which the algorithm can design. This is where the role of AI in design struck me. In less than two days the neural network created more than 30.000 unique chairs, something that is impossible for a human designer. 

The possibilities of AI are unlimited but we as designers and engineers can make sense of it by giving it boundaries and curating the results. It opens up possibilities that go beyond what is capable for humans in terms of creativity, scale and speed. We at Radix believe that the role of a designer working with AI solutions will evolve more into the role of a curator i.e., choosing which outcomes make the most sense. This act of sense-making is really appealing to us and we think this is what design is about. After all the word design comes from the latin word ‘designare’ which translates to ‘appoint something’ and thus give meaning to something.