AI, Sustainability, and Design — A Call to Action
The 2022 Sustainability Development Progress Report is not fun reading. The report is summarized in seven alarming key messages. All of them points to lack of progress and bleak prospects for the future…
One of the key messages, however, has a slightly more positive tone. It states:
“Key Message 06. Data and statistics can light the way. The report also shows that we still lack timely, high-quality and disaggregate data to fully understand where we are and where we are headed. Investment in data and information infrastructure should be a priority of national governments and the international community.” (The UN, 2022)
Data science and well-designed AI-powered services that can explain and motivate us to act thus play a crucial role when we address the sustainability development goals. But as it turns out, AI may also inhibit some sustainability development (Vinuesa et al., 2020). This makes adopting AI for sustainability a wicked problem. There are several interconnected dependencies, and as soon as we start addressing one part of the problem, the complete problem changes. There is no stopping rule, and we only have one shot at getting this right. This requires prototyping, speculative design, and efficient iteration of solutions. In fact, wicked problems are best approached with design and systems thinking methods (Farrell & Hooker, 2013).
In order for AI to be useful, it needs to augment the innate human superpower of judgement and her ability to act. This requires thoughtful interaction design and service design. Furthermore, as AI services become more intelligent, they need to rely on well-designed human-agent interaction (Johnson & Vera, 2019). Visualization and design for human understandability makes human-AI interaction more efficient. In fact, it’s the most crucial aspect:
”Research in visual analytics and human-centered data science shows that one of the most important elements for maximizing the effectiveness of an algorithm that is designed for humans to use is transparency or understandability.” (Aragon et al., 2022)
Communicating and visualizing complex problems to aid human judgment and action is crucial. Humankind’s most efficient way to understand the world, and complex patterns, is through story. We are, by evolution, “storytelling animals” (Haven, 2007; Gottschall, 2013). Storytelling, visualization, prototyping, and human-AI interaction are all in the toolbox of humanity-centered design.
Therefore: addressing climate change, biodiversity, and other sustainability challenges (our time’s most important problems) requires human-centered data science and AI (humankind’s most advanced technology), strong visualization and storytelling for understanding, and that we view the interplay between AI development and sustainability goals as a humanity-centered and planet-centered wicked design problem (Wärnestål, 2022).
I don’t think there can be a clearer and more epic goal for our profession. Data scientists, designers, and AI engineers: assemble! The world needs you as heroes in the end game.
- Aragon, C., Guha, S., Kogan, M., Muller, M., & Neff, G. (2022). Human-Centered Data Science: An Introduction. MIT Press.
- Farrell, Robert, and Cliff Hooker. 2013. “Design, Science and Wicked Problems.” Design Studies 34 (6): 681–705.
- Gottschall, J. (2013). The Storytelling Animal: How Stories Make Us Human. Mariner Books.
- Haven, Kendall (2007). Story Proof — The Science Behind the Startling Power of Story. Libraries Unlimited.
- Johnson, Matthew, and Alonso Vera. 2019. “No AI Is an Island: The Case for Teaming Intelligence.” AI Magazine 40 (1): 16–28.
- The UN (2022). The Sustainable Development Goals Report 2022. https://unstats.un.org/sdgs/report/2022/
- Vinuesa, Ricardo, Hossein Azizpour, Iolanda Leite, Madeline Balaam, Virginia Dignum, Sami Domisch, Anna Felländer, Simone Daniela Langhans, Max Tegmark, and Francesco Fuso Nerini. 2020. “The Role of Artificial Intelligence in Achieving the Sustainable Development Goals.” Nature Communications 11 (1): 1–10.
- Wärnestål, Pontus. 2022. Designing AI-Powered Services. Lund: Studentlitteratur.