Everyone can remember that one scene in a movie where a computer seemed magical. From the enormous, button-blinking desk consoles of Star Trek to the rapid-fire typing of Dennis Nedry hacking into the mainframe in Jurassic Park, dashboards are met with a sense of mystery and trepidation. This mystery continues even in today’s companies. One reason for this is that traditionally, data scientists and user-experience designers have not worked side-by-side. Many of the machine learning models behind AI were developed to sort, order and analyze huge swaths of data. They were built only to supply an outcome – rather than an easy, intuitive user experience.
The widespread adoption of user-centric software – in the form of our favorite apps, email services, and even fitness programs – has turned elegantly designed and easy-to-use AI models into an expectation. Programs with a seamless user experience are limitless in their potential to improve human interaction with technology and can be applied to everyday tasks within enterprise businesses – from managing and utilizing financial data for improved efficiency, to optimizing production to ensure product quality and consistency.
It is essential developers go beyond algorithms to actually build a beautiful, easily operated end product to encourage adoption among technical and non-technical users alike. That’s the future of software – no matter how automated, there is always going to be a moment a human being needs to use it. Here are a few ways organizations can start designing AI and ML models with a user-centric approach:
Get your designer and your data scientist in the same room
While data scientists and designers typically do not overlap, creating human-centered models of AI systems requires a delicate marriage of the technical (scientists) and the functional (designers).
Developing through the dual lenses of user-centric design and machine intelligence is how organizations can, for example, refine a united control panel to manage AI against all business outcomes, enabling executives and marketing teams – not just data scientists – to easily interact with the system.
Computers are equipped at analyzing enormous quantities of data, so it makes sense to do away with complexity and move towards systems that consolidate information, presenting only what is necessary. Together, data scientists and UX designers can identify user pain points and work towards a unified model that goes beyond information and insights, but also improves the manner in which this data is presented to better meet the user’s needs.
Building more elegant solutions is also essential for scaling the technology to handle more without needing a highly trained operator. The more the systems can do, the more a human can manage.
Copy and make it better
According to a study from McKinsey on the State of AI in 2020, 60% of respondents have a clearly defined AI vision and strategy – but how will these businesses ensure the implementation of these AI models are functional for users?
A key component to creating AI systems for human use is to build and improve upon previous models. UX is an iterative process that requires trial and error. That means conducting research on how users have historically responded to these systems and collecting feedback from tests, identifying areas for improvement, and executing a careful process of implementing changes to perfect the application’s design for users. In fact, research from acclaimed HCI expert and usability researcher Jakob Nielsen, shows that reworking an interface can improve usability by a median rate of 38% per iteration, with the overall process having the potential to improve usability by a full 165%.
Finding that balance between functional and informative systems means seeing what has worked in the past and how it can be improved. In addition to internal research, organizations can do a competitive analysis and an audit of their market to see who is creating intelligent solutions and find ways to incorporate good ideas into their own work, in a manner that works best for them.
Education is everything
If your designers and product owners do not understand different AI techniques (e.g., computer vision, NLP, RPA) then they aren’t going to know how to use them. Therefore, it is essential for product teams to align on a specific mission, become familiar with each other’s roles, and understand the various applications of AI. To get teams in a place where they are agile, consider implementing regular training, seminars, and educational meetings on aspects of both AI and UX and how they relate to the final product for the user.
A focus on education ensures teams are beginning with the end-user in mind – which means understanding what AI-powered systems are designed to accomplish as well as the best way to have that information presented. Spending time on this upfront results in fewer misunderstandings and revisions in the product development process, so teams create better solutions that go to market sooner.
I cringe thinking of the MS-DOS operating systems I still regularly encounter. It is crucial for digital companies to build programs with strong UX/UI and embrace a hybrid scientist-designer approach in order to remain competitive. The potential of AI is exponential and human-centered design will further drive growth by acquiring and retaining users through better experiences.
By Chris Klee, EVP of Design at Hypergiant
