It’s no secret that artificial intelligence (AI) is a hot and trendy topic right now. AI is so popular that it’s been drawing the attention of people from various disciplines. With all kinds of professionals looking at the topic and its applications, a geeky designer like myself wasn't going to be left out.
Although AI’s popularity in the mainstream is rather new, my journey down this rabbit hole started with a passion for technology when I was very young.
The digital world was changing rapidly as I was graduating from design school. Digital products were pushing their way through and terms like user experience (UX) and user interfaces (UX) and their respective fields were gaining traction. The UX/UI discipline was exciting and appeared to be a perfect place for newly minted design graduates to get their first jobs, and so I did just that. Right out of design school, I took on a role as a UX/UI consultant for a software company.
While at this software consultancy firm, I started to notice that big data and bots were some of the concepts being used to convince clients to develop their systems. At the time, I had little idea of how important and widespread the use of both of those concepts would go.
A couple of years and a few digital companies later, I landed a job at an autonomous delivery startup. This new position was at first mind-blowing. Getting to see robotics (a robotic courier) come to life was a delightful experience and, as a product designer, going back to tangible artifacts as products was inspiring.
Although the startup hired me to be part of their logistics function, I had the opportunity to jump to the manufacturing and maintenance functions partly, which I took because of the thrill of designing actual physical products. Of course, loving the design aspect wasn't enough to satiate my appetite for learning how these things work. Understanding how the little robotic courier worked was also a crucial part of my job, which furthered my interest in AI.
( Regarding how robotic couriers work, they work partly due to computer vision and machine learning (ML). For those not familiar with autonomous vehicles, the idea is to make the car learn how to drive. Although in reality, it's more like teaching the vehicle how to know where it is, to avoid obstacles and to slow down.)
One of the endeavors I was tasked with while designing the product was to determine how to place the cameras on the robot so that the data science team could gather enough information to "train a model." Upon hearing this, I then asked the obvious question: well, how do you train a model?
Thanks to this question and a friend of mine, David, I decided to give finding an answer a go. With David leading the training, we began by going over some high school math concepts and evolved from there to data analysis and, finally, machine learning.
Moving from being strictly a designer to a designer who also plays in the data science and machine learning worlds came with its challenges. Some of the challenges included:
It is common for designers to avoid mathematics while in college or at university. But when it comes to learning machine learning, it’s a good idea to go back and grab a few math books to learn the fundamentals. Understanding the underlying mathematical concepts will speed up the learning process when it comes to learning machine learning and will make the process much less frustrating. MIT 101 Statistics and probability are pretty good for this.
It is always a good practice to ask questions about things you don’t know to people that do know. I know it sounds like common sense, but, in this case, there’s a trick to this:
Be persistent when asking data scientists about machine learning concepts. In the beginning, most of them will try to over-simplify machine learning for you. Although that’s good for early stages, eventually, you will need to step up your game. So when it comes to getting concepts out of them, I find it especially useful to use their language, be specific and not leave the room without having a clear answer!
One dimension of design is the craft of abstracting concepts to make them more usable for users. So the question of “how” arises when your task is to make the already abstract into something more abstract yet understandable and, more importantly: valuable. This question brings on a whole new domain of design challenges.
One of the most valuable skills that design can bring to the machine learning world is: Data Visualization.
Designing rich data visualization experiences becomes extremely important when explaining what happens with specific data in a machine learning model. It requires context and scale. This provides a perfect spot for graphic designers to be a bridge between DS/ML experts and non-experts alike thereby allowing this new technological revolution and its applications to reach further than it already has, empowering through explanation.
As technology evolves so does the design profession. Unveiling the mystique behind data science and, especially machine learning, is a new design challenge that designers in the machine learning more like myself are facing.
To precisely represent a machine in the act of learning and predicting will help us squeeze out the full potential of AI and take it to places where it has not been yet. Can you think of more ways designers can add value to the machine learning industry?
Project manager and Design Thinker with iterative and creative logic. A believer in design and leadership as creative disciplines, well structured and with methodologies, separated from the notion of design as a compendium of techniques and leadership as a talent. Currently making Machine Learning explainable.