The hands of a woman holding an ipad. She is pointing on the screen while a robot hand is also pointing.

Shaping new tools: Machine learning in predictive design


Artificial Intelligence (AI) is about to change the discipline of design. But with great potential comes great responsibility. While AI has the potential to reshape the way we work in design, we have to carefully define when and how to use it.

Machine learning and design tools—the next step in the evolution of design

Wouldn’t it be nice to know how people will look at your design? Using machine learning, new design tools show how this might be possible. Predictive design is about using past data and statistical methods to predict patterns in a design. For example, a system might take data from previous attention map tests that show where people look in a design. From that, the system then anticipates where people are likely to look in other designs. The more data is fed into the machine learning process, the more accurate the predictions become. This can save us as designers time and effort in research, with AI taking over work from us.

Currently, there are a number of design applications out there that have adopted such non-human evaluation systems. Although it is too early to draw conclusions, these systems hint at where predictive design is headed in the years to come. A concrete example is Visual Eyes, which uses data from real eye tracking tests to predict where users will look. Visual Eyes then creates a heat map that shows these predictions. Bonus: it offers plugins for Figma, Sketch, and Adobe XD, and can thus easily be integrated with existing design tools. This way, the AI element helps to make the workflow of designers leaner and more efficient.

A heat map test done with Visual Eyes on our Peerdom Design showing the most looked at areas in colours ranging from red to blue.
A heat map test done with Visual Eyes on our Peerdom design.

Making a useful tool for humanity

AI has the potential to change the way we work as designers. But the use does not only bear advantages; there are risks as well. For example, decisions taken during data collection can seriously impact the output. Questions like “Whose data do we include and whose do we exclude?” are something machine learning does not automatically take into account. Let’s go back to the heat map example: While the heat map gives insights about aesthetics, it does not automatically favour a more inclusive design. Someone with a disability might seriously struggle with what the algorithm identifies as “optimal design”. These are concerns that we as designers still have to keep in mind when making use of those tools. So how can we keep things human and still profit from machine learning?

Although it might be convenient to “take the human out of the loop”, it is our mission as designers to find out where these machine interventions can help to solve human problems. This way, machine learning enables us to focus our energy and attention on the creative and emphatic side of things. At Nothing, with increased use of AI in our own work, we also make sure to put a strong emphasis on the humane. It is our goal to understand people so that we can then use technology to empower everyone to go beyond their current capabilities.

More material on AI and human interaction:

We’re excited to exchange about the potential and pitfalls of applying machine learning to design. If you have a specific question where you’d like our input, talk to our expert in the latest machine learning tools, Nathan aka Pace (