Design Doing with AI
Artificially intelligent systems are useless without data. Collecting data without a clear goal is pointless. In order to actually generate value, data needs to be leveraged into activities by AI – otherwise, our quest to productively unite human and machine becomes hopeless.
Men and machine are already working together successfully in the medical field. Studies show that the cooperation between physicians and AI leads to significantly better results than if each party diagnoses, operates and works on his own. But how do you design the collaboration between humans and AI, how do we develop purpose and objective in a system where man manages the AI?
The common opinion in many companies: A project with AI is an IT project. Because it needs processed data, appropriate computer capacities and deep learning algorithms. All things traditionally attributed to IT.
A fallacy, which I often experience, and which we lapse until we finally learn to work silo-overarching. In terms of AI, for example, there are no learned use cases. Based on real-time data and continuously new insights, we will be able to address any number of variable and complex scenarios, increasing the granularity of the use cases to infinity. Even personas become irrelevant. Each individual user becomes a persona-self based on one’s individual data track.
Also, “real AI projects” are not really foreseeable or predictable. Neural networks are probabilistic and heuristic. So, it’s about probabilistic calculus. You do not know in advance what you will learn, how far you can go and what you want, or whether the result is good at all. Just as unpredictable are the costs. This is a horror for the classic IT manager and, in terms of complexity, these project cannot be managed from one department alone.
And now? Are AI projects equivalent to fighting the Hydra or THE professional adventure of our time?
If we want to tackle AI projects successfully, we need a corresponding mindset and suitable tools. Interdisciplinary people with a focus on the user must be part of the team. Important: AI is a tool in the project and key technology of the solution – but never the result. As always, you should start with the right questions first. That’s why the five basic steps of the design thinking process help here as well. With the difference that at the process and method level data and algorithms participate integratively.
Identify:
Here man tries to understand the human. His perspective, problems and situation. On the one hand through the classic methods such as shadowing, on the other hand with the help of AI, by gaining relevant and new insights and patterns from the gigantic sea of data in real time. Otherwise this was hardly possible due to the amount of data and the missing context.
Explore:
In this phase the challenges and opportunities are defined. AI supports the project participants in the prioritization of these. And yes, as a key technology, AI can also be part of the opportunity landscape.
Ideate:
Finally, humans and AI can go hunting for great ideas together. Every brainstorming experience is powered by an AI-driven idea suggestion system, which is fed from all kinds of data, a firework of creativity and speed. Also, the real-time evaluation of ideas (e.g., matching patent database) as well as the prioritization based on user preferences (user data) is possible.
Create:
Even if AI is not part of the key technology. In this step, it helps to develop alternative design solutions and prototypes and to test them immediately. Wherever there was no time for iterations, AI allows multiple passes in the shortest possible time and with the highest quality.
Scale:
The MVP is now entering the market. Again, AI helps us to determine which market is likely to scale most successfully. Continuous tracking and small optimizations are made faster and more targeted.
Each of the five steps will have their own new tools like AI Interaction Play, AI Empathy Testing and many more. Especially if AI should work as part of the solution. Because not only do the participants of the project have to learn how to collaborate with AI, but the end user must also accept AI as part of the solution. The worm has to taste good to the fish, not the fisherman.
AI is not (just) an IT project. But must have a benefit for the people. As the saying goes, whoever has a hammer as a tool sees nails everywhere. Therefore, one should only then decide on a technology or platform (Chat Bot or Conversational App, etc.), if one really know what one want to achieve. Another reason why Design Thinking is my preferred method, at least for now.
First published by PAGE online in German.
Karel J. Golta
CEO + Founder
