A designer identifies user problems or opportunities and creates desirable, viable and feasible solutions. Recently there have been questions like “How could we use AI, machine learning or GPT3 for something useful?” This leads to a technology-driven project, which is fundamentally different from the typical design process.
I used to work at Nokia, a marvellous technology company. Nokia once turned mobile communication technology to a consumer product and made it desirable, so that most people in the world bought a mobile phone. Yet, at its heart, Nokia was — and still is — a technology company. I worked with researchers and engineers who developed new technologies. Thanks to this, Nokia still has a vast portfolio of technology patents.
Nokia was excellent at solving engineering problems, but much less so in knowing which problems to solve. One of the typical project types that we had in those days was to come up with ideas, what a new technology could be good for. What is the opportunity or problem that a novel technology could solve? From a designer’s perspective, this is a particularly difficult type of design: finding problems for solutions.
This is also currently our challenge with AI. It is a solution looking for a problem. I’m by no means saying that AI would not have useful applications — it certainly does. The designer’s challenge is that if the goal is to find use for AI, it starts from the solution instead of the problem, and this is not how designers typically work.
There are almost as many design processes as there are designers. However, there are some models that are quite prevalent in the field of design.
A typical design process (see [1]) starts off with an analysis phase: we want first to observe the world of the user or the customer: their wishes and doubts, their current use patterns, and their physical, cognitive and social context of use. Based on that we can analyze, what their underlying needs and motivations are. Only then we know what kind of solutions will match these needs. At the same time we need to make sure that the solutions are implementable and provide a reasonable business model.
In this process, we always first understand the user group and their needs, and only then select whichever technology would be the most appropriate to fulfill that need. As you can see, even if we had a particular technology in mind before the exploration, we may find out that some other technology (or possibly no technology at all) matches the user needs better.
(Sidenote: despite being popular, this process model is not how design is done in real life. It is much more agile. Read more e.g. [2])
How to find good use for technology?
The standard design process doesn’t work when we want to find a good use for a new technology. If we already have a solution, the question is, how can we find a problem, for which this solution would be the most optimal?
First, it has to be noted that many successful innovations have been created by coming up with the technical solution first, and only later have people found really useful applications for it. Just with sheer luck, the first use case is a hit. Sometimes the perfect match is found through somewhat chaotic evolution: the technology is first used in an application for a completely different purpose but then later adopted for another use that was in need of a solution. The survival bias amplifies this: we only remember the successful cases. The failed attempts that have had only little exposure are forgotten quickly.
One good example is text messaging. It was initially considered a technical add-on for cellular network operators to send one-way messages to mobile phone customers, for example about received voice messages. Luckily, the standardization committee added an option for “mobile originated” messages, which meant that mobile phone users can send short texts to each other, not just receive them. This certainly hit an unmet need, and before we knew it, the whole world was texting.
Technology-driven design process
So far, I have found only one way to do technology-driven design: by doing an exhaustive search. The process consists of the following steps:
- Understand the potential of the technology: what can it really do.
- List all different potential usage for the technology that you can come up with. Identify all potential target groups who could be interested in it.
- Prioritize the hypotheses and select the most promising ones for the next steps.
- For each selected hypothesis, go through the design process: analyze the needs of the intended user group and check if the technology is really the best fit for this need.
- If you find out that the technology is not the best option for that user’s need, you must abandon this hypothesis and move to the next one.
As you can see, this process can explode: there are often too many potential uses for a new technology. For each user group, analysing the user needs and iterating the design and the solution takes a fair amount of effort.
A technology focused design process is significantly more time-consuming and costly compared to traditional user-centered design project.
The key for keeping the technology-focused design process manageable is that we try to identify and work on only the most potential hypotheses — user groups and their expected user needs — and select these to the next phases. If any of the readers have further ideas or methods for doing this prioritization, please do share e.g in the comments!
Coming up: AI in designer’s toolbox
From a designer’s point of view, AI is still very much a solution looking for a problem. For further ideas, how designers can learn how to include AI solutions in their designs, tune in to the next blog post (coming soon).
[1] “Five Phases of the Design Thinking Process”, Interaction Design Foundation
[2] “Re-thinking design thinking”, Panu Korhonen, 2019
[3] “How to Find Real Use Cases for a New Technology”, Stellex Group
[4] “How to find winning artificial intelligence use cases”, ComputerWeek