Adobe’s Machine Intelligence Team designs the features behind the artificial intelligence and machine learning functionalities of its creative software. When design lead Patrick Hebron visited the MFA Interaction Design studio, he addressed how machine learning systems play a role in different features of a product and what the implication of new technologies can be for user-facing design.
Visual explanation of a simple machine learning system that is governed by multiple variables from Patrick Hebron’s book, Machine Learning for Designers. (Image: O’Reilly Media
To explain the core logic of a machine learning system, Hebron compared the system to a coding language. Code has a formal structure. It’s boolean; there’s a true and a false. Machine learning is a bit fuzzier, not quite black and white. Machine learning systems learn from a range of experience and data. They are prone to errors, especially if the desired outcome is a concrete series of actions that will produce a desired functionality. Machine learning systems learn from context. With each additional experience comes better mapping and an increased understanding of the world.
Accordingly, the design opportunities for machine learning systems are ones that can parse complex information, like deciphering human facial expressions or allowing for multimodal input. At Adobe, the Machine Intelligence Design Team sees this as a tremendous opportunity to transform the ways in which we communicate. One example, Hebron explained, is advancing machine understanding of natural human behavior. Through machine learning systems, we can start to design approaches to help users communicate their ideas in ways that a machine can quickly and accurately understand, or even auto-complete an action using human input of language or handwriting. People can use machine learning systems to craft and articulate ideas in the mode that is most powerful for them, expanding the way we all share ideas and express creativity.
Adobe After Effects’ content-aware tool allows you to remove unwanted objects from videos. (Image: Adobe.
Hebron’s team builds tools that use machine learning systems to enable creatives to do more and think bigger. As machine learning makes using creative tools more organic to human thinking, their accessibility opens. By improving existing functionality and adding new components to these tools, creatives can test the boundaries of creative exploration and expression.
As creative professionals designing for new technology, we must communicate its limitations in addition to its opportunities. “The cutting edge is serrated,” Hebron said. “Design is needed to even out the surface and fill in the gaps and make a straightedge out of a serrated one.”
Follow Patrick Hebron on Twitter for updates on his work.
