The Need for AiX Design - Part 1

Digital Product Design in the Age of Artificial Intelligence

Considering all the hype and high hopes surrounding the progressing integration of artificial intelligence in consumer and business products, too many product development efforts still seem not to deliver on the value promised. While over-ambition and unrealistic expectations, fueled by the never-ending stream of exciting academic achievements in the field, can play a part in this, there is another issue at play as well – an issue of understanding. Traditionally, where the language of technology is precise and concrete the language of business is anecdotal and narrative – a frustrating communication gap addressed by inventing design processes and design systems, among other tools. However, since the emergence of AI, proven best practices do not suffice anymore, and a hands-off attitude by product managers, subject matter experts or designers will just not cut it. We need to relearn how to conceive and design digital products enabled by AI-technologies.

Aspirational goals need to be translated into cost functions. Computational results are now only somewhat correct and sometimes utterly wrong – with no bugs to blame. Data is at the center of everything, and all endeavors become big experiments. On the other hand, there is the opportunity to create truly innovative tools and services.

Should tried-and-true methodologies be abandoned? Certainly not. A successful product still needs to address a job-to-be-done by people (in common parlance, a “user need”). Understanding how people go about their tasks, what challenges they face, and what aspirations they have, is still crucial. However, pretending nothing has fundamentally changed and treating AI technologies as an add-on in established product design processes would be naïve. Understanding the solution-space continues to inform the problems identified to be solved.

This is not to say that product managers or designers must learn Python and how to train their own models, but just like the architect must understand the peculiarities of the building materials and environmental conditions they plan for, there needs to be an intimate understanding of the consequences of the technologies that make our products run.

Treating AI-technologies as magical is fine – treating them as mysterious is not.

Companies in which AI development efforts are chiefly technology-led deliver value only half the time, compared to ones with a broader understanding (Ransbotham 2019, MITSloan Management Review). Not only do companies need to build technological capabilities but they also need to ensure that enough internal business users have the ability and desire to consume those capabilities to translate them into real customer value.

Some of the big players and some independent initiatives are starting efforts to articulate the opportunities and challenges attached to AI-technologies in a business and design relevant manner, for example Spotify, IBM, Google, Futurice, AI x Design and others. These are important first steps.

There is no room for complacency with only understanding the fundamentals of the technology. As product designers, product managers, and subject matter experts, there needs to be a proactive role in setting up new methods, processes, tools, and best practices to exploit the potential AI-technologies’ offer; this will enable the development of products that are not only faster, cheaper, or better, but enable their users in new ways all together.

Black-box thinking might get us past horizon 1, maybe horizon 2, but will keep us from ideating and executing truly transformative product concepts. Amazon could stop at using AI for product suggestions or at predicting demand to distribute products across warehouses for fast and efficient delivery. Instead, reliable predictions could as well flip the fundamental sales paradigm from buy-then-ship to ship-then-buy, sending customers products for purchase, being confident that only a few are sent back, which increases revenue and customer satisfaction. In addition, black-box thinking is being challenged commercially and ethically. Explainable AI is emerging as a business requirement in healthcare and related industries, where models need to be explained to build trust, maintain quality and confidence, in the solutions we build.

McKinsey (2017) predicts that half of all work activities world-wide have the potential to be automated by 2030, while 60 percent of occupations might see at least 30 percent of their activities automated. 8–9 percent of the population will work in occupations that do not exist today. Productive work will not decrease but social and emotional skills, management, creativity, and judgment will increase in importance. Maybe surprisingly, automation will often demand even better skills from its human collaborators. We can expect AI-technologies to have a similar impact on private and social life as well.

To discover promising product concepts and business models we need investigate which tasks and to what degree automation is most valuable. AI makes up only one half of the human-machine collaboration. Finding synergies, considering human psychology of adopting technologies, the impact on people’s work and lives, ethical implications, and so on, raise questions that are not to be answered by technologists, but by product managers, designers, and user researchers.

Now is the right time to invest in AiX Design!

Want to read more? Read part 2 of the AiX Design System series.

Useful reads

To get a feel for some aspects of the topic, I recommend these two books:

From a design perspective: “Designing Agentive Technology: AI That Works for People” by IBM’s Christopher Noessel.

From a business perspective:“Prediction Machines: The Simple Economics of Artificial Intelligence” (available in a neat bundle at HBR)

Or if you prefer video and want to get a perspective on AI-ethics:“How can machines learn human values?” by Brian Christian

References:

Manyika J, et a., 2017. ‘Jobs Lost, Jobs Gained: Workforce Transitions in a Time of Automation’, McKinsey Global Institute, McKinsey & Company.

Moldoveanu M, 2019. ‘Why AI Underperforms and What Companies Can Do About It’, Harvard Business Review, Harvard Business School Publishing Corporation.

Ransbotham S, et al., 2019. ‘Winning with AI – Pioneers Combine Strategy, Organizational Behavior and Technology’, MITSloan Management Review, MIT.