What Designers Need To Know About Algorithms
As almost everything around you has gone through a design process, algorithms need some design 💖 #too. I’ve been working with data scientists intensively over two years now, trying to bring both worlds a bit closer together.
So Happy Together 👩🎨👨🔬
Imagine a data scientist and you? I do! I think about it day and night.. 🎵
As we, designers, primarily focus on qualitative research in the first design phase, the pace for data scientist differs since they’re focusing on quantitative research. As designers, we can wrap up this research in a certain time period, or design sprint, where data scientists can hardly predict how much time they need to validate the riskiest assumption.
To the rescue!
I’ve seen that helping out in this research part from a design perspective really makes a difference in efficiency and algorithm quality. If the key feature of your product or service is in the science why not try to tackle this assumption asap together?
Transparency is key 🔑
With algorithms, we try to recreate and shift human intelligence and memory to a computing process. For this, we need valuable user input, but also an idea of how they work and why they do it. Sounds a lot like UX research, right? If you want to get these insights from meaningful conversations and you want to be able to read between the lines, transparency is the key to gain trust.
Code that learns is powerful but dangerous. It threatens the rules of markets and civic life. AI requires a new technical and civic infrastructure, a new way to do business, a new way to live together in communities.
To avoid building too much of a black box, it is best to be as clear as possible about what you want to build and why.
The key takeaway of the popular ‘Hooked’ book by Nir Eyal is that we shouldn’t design things where the user is missing the knowledge you have. This also applies to openness in developing algorithms. So here is the how-to.
Use ‘stakeholder-readable’ flowcharts
Although algorithms as part of AI are easily interpreted as rocket science, most of the times they are not. You can see an algorithm as a mathematical function or a set of rules that lead to a certain result.
After discussing why you want to co-create, flowcharts are a great way to go about creating an algorithm and give an understanding of what you are trying to achieve and how. Is this any different than a so-called low-fi prototype?
If you can’t explain it simply, you don’t understand it well enough.
This way you can discuss with your stakeholders what they think about your model and iterate on the design. You can specify where it can possibly lead to and how it will influence their way of working. Just telling them how it works instead of showing, isn’t enough.
What are you sayin’?
Pair up with the geek, create a transparent box of data science goodieness and let your users and stakeholders help build it.