The world’s leading publication for data science, AI, and ML professionals.

Merging Science and Design to Make Artificial Intelligence for Everyone

Design's human touch can help bring AI design into everyone's reach.

Image via Unsplash
The more involved we are in building AI systems, the less intimidating they’re going to seem, and the less destructive they’re likely to be. Photo by Possessed Photography on Unsplash.

By now, everyone’s heard of Artificial Intelligence (AI) and its little cousin, Machine Learning. AI’s various sub-fields, like machine learning, computer vision, and so on, are perhaps one of the most used and yet least understood technologies out there today.

In fact, many people who work in the field actually call AI systems "black boxes" because they can see the inputs and the outputs, but they don’t actually understand what’s going on inside of them. This would be troubling enough for some experimental tech in a lab somewhere, but just think about the fact that many of these machine learning systems are responsible for making decisions that affect numerous people’s lives.

Who really knows what's inside, or what makes an AI tick? Photo by Sam Moghadam Khamseh on Unsplash.
Who really knows what’s inside, or what makes an AI tick? Photo by Sam Moghadam Khamseh on Unsplash.

Just some of these decisions include:

Yikes right?!

It’s not all doom and gloom though. More recently, there have been efforts from the tech & science community and the design community to make AI systems more explainable and transparent. In other words, to allow people to understand the decision-making processes that happen inside these systems and how the input data gets analyzed and affects the outcome. While these efforts are definitely a step in the right direction, they are often after-the-fact, responsive interventions. Instead, preventative measures might be more valuable, ensuring that these systems are built from the get-go using insights from the people who will be affected most by them.

Tech-based and design-led interventions are helping to shed light on the inner workings of black box AI systems. Photo by Hassan OUAJBIR on Unsplash.
Tech-based and design-led interventions are helping to shed light on the inner workings of black box AI systems. Photo by Hassan OUAJBIR on Unsplash.

Cue: Co-Design

Collaborative Design, or co-design for short, is the act of designing with people, as opposed to the traditional designing for people. Gaining popularity in Scandinavian design practices at first, this approach to design has gained immense popularity in recent years because of the value it brings to the design process.

By involving users and other stakeholders who are going to be affected by whatever is being designed, the design team can understand their needs, opinions, and experiences very early on in the process. This helps the team factor in this information from the get-go, as opposed to building something and then finding out that it’s unsuitable during the final testing phases, which is when users and stakeholders would traditionally have gotten involved. Another benefit of co-design, as opposed to just interviewing or surveying users, is that it helps overcome what’s known as the "stickiness" of users’ knowledge: i.e. the difficulty users have in actually saying what they need. By engaging them in fun, interactive design sessions, participants tend to feel more comfortable and have different channels to express themselves beyond verbal explanations.

Co-Design can unlock user information and experiences through non-verbal channels. Photo by UX Indonesia on Unsplash.
Co-Design can unlock user information and experiences through non-verbal channels. Photo by UX Indonesia on Unsplash.

Science and Design: Two Sides of the Same Coin

You might be thinking: how can we seriously bring in a user who has zero technical knowledge and ask them to design an AI system that most people with backgrounds in tech struggle to understand?

And the truth is, it’s easier than you think! And it’s been done before.

There is a constantly increasing number of studies that have focused on applying different co-design practices to AI design processes. These studies have used techniques like role-playing and decks of cards to help different non-stakeholders design and make decisions on the different features and behaviors of AI systems.

Using these techniques and involving more stakeholders has numerous benefits:

  1. It leverages "community expertise" to make products more empathetic and human-centered, ultimately increasing user acceptance, trust, and buy-in.
  2. It increases inclusiveness and participation when it comes to key decisions being made, potentially leading to less biased, narrow-sighted outcomes with dangerous implications against certain groups of people.
  3. It allows for more a interdisciplinary set of people working on designing products, which has countless benefits – and I’ll talk more about this down the line.

In fact, several branches of design are extremely valuable to tech projects and especially AI systems, and I’ll be introducing those branches and their value in another article later on.

The Way Forward

While existing projects applying design techniques to the field of tech, and specifically AI, are beneficial, they’re not really enough to have a far-reaching impact yet. A lot of these efforts have been isolated, individual studies whose results tend to not be used in other applications or on wider scales. These projects and several companies have taken great strides in the right direction, but what is missing now is:

  • To replicate and test studies across domains and applications to see how well they generalize,
  • To create a full methodology or process that formalizes and standardizes the inclusion of different stakeholders across the entire AI life-cycle and not just in initial ideation phases,
  • To focus on the values that matter most to stakeholders and how to respect and uphold those in the technology being created.

There’s definitely promise in applying co-design to creating AI systems, but next steps need to be taken in order to mature the practice.

New and exciting work is taking place at the intersection of design and AI, but there is still a long road ahead. Photo by Jukan Tateisi on Unsplash.
New and exciting work is taking place at the intersection of design and AI, but there is still a long road ahead. Photo by Jukan Tateisi on Unsplash.

Where I Fit In

This current reality where AI systems hang in the balance with the potential to become even more isolated, exclusive and complicated; or open up and become more accessible and inclusive, is what inspired my PhD project. By looking at points #2 and #3 above, I’m working towards creating this unified process, and a toolkit to support it, to systematically involve people throughout the AI life-cycle – with a focus on value-sensitivity.

You can check out the official page for my project on the Imperial College London website. You can also check out this other article I wrote explaining the details of my PhD project.

I’ve set up this Medium account to publish interesting findings as I work on my PhD project to hopefully spread news and information about AI systems in a way that makes it understandable to anyone and everyone. This article is the first of many I have planned to explain a variety of concepts, as well as some updates on workshops I’ve already run and some cool topics from my literature review. If you’ve liked this first article then please consider following along as I post new things, and please like and share!


Related Articles