Artificial Intelligence: A Superpower for Human-Centred Designers
One day the AIs are going to look back on us the same way we look at fossil skeletons on the plains of Africa… an upright ape living in dust with crude language and tools, all set for extinction.
While we’re still far from the realization of this quote from 2014’s Ex Machina, the capabilities of artificial intelligence systems have grown exponentially in the past few years. From predicting election results and drone navigation; to diagnosing cancer and scheduling meetings; these are just a small sample applications of AI that will revolutionize how the world works in the upcoming decade, and beyond.
http://medicalfuturist.com/wp-content/uploads/2015/02/Artificial-Intelligence.jpg
Design in the Exponential Age
Industries and institutions across the globe are being disrupted. The rate of change in technology and society has flipped previously indisputable truths on their heads. From the rise of Netflix over the TV industry to the success of Brexit and election of Donald Trump, “expert” opinion has proven to be fallible. Businesses looking to succeed in an age of exponential growth must look for new ways to find their niche. Companies like Facebook, Apple, and Tesla are leading the way in their respective industries (social media, consumer tech, and automotive/energy) by delivering a service that seamlessly integrates quality products (video, phone/tablet, car/batteries) with quality user experiences.
In hopes of imitating the success of these technology titans, companies are seeking out service design or innovation consultancies (from now on referred to as design consultancies) to help them identify the secret sauce for creating value for themselves and their customers. By applying a concept called “Human Centred Design” (HCD) into their practices, these individuals or organizations are able to ignite innovation through the generation of unique insights, engagement of key stakeholders, and transformation of their client’s strategy and business model. For these firms, making human life better is the goal, and Artificial Intelligence (AI) is a tool, not the solution [1]. Innovation is now table stakes, and the integration of HCD & AI is the winning play.
The Jobs of a Design Consultancy
A design consultancy’s value lies in its ability to uncover deep human insights and translate them into strategies and tactics that solve a burning unmet need. While each firm has their own combination of methods and tools to support their clients, the underlying approach can be summarized as three key “jobs”:
Research
Translation
Facilitation
(These jobs can all be performed by a single consultant or design strategist, or shared across diverse teams of human-centred designers)
Research typically takes the form of ethnographic interviews, observations, and/or immersion experiences that involves trying to understand how the client’s key stakeholders live their lives [2]. Because informants (those being researched) are often unaware of their own needs and preferences, the consultant is more than a data collector, rather they are someone who could elicit meaningful stories and behaviours across a broad scope of activities — like a partner or confidant. The output of this research can take the form of voice recordings, transcripts, photos, videos, and artifacts provided by the informant.
With data in hand, the next job is Translation, which entails structuring, interpreting, rearranging, and reporting on the data collected from stakeholder research. Finding patterns and making connections is equal parts art and science, and takes experience and training to improve. As there are often no clear right or wrong answers, it is sometimes necessary to go back into the field to check assumptions and validate insights — iteration is inherent in the human-centred design (HCD) process. The output can range from a simple list of universal needs or design criteria, to grand visualizations in the form of journey maps and service blueprints.
The final key job is Facilitation, involving the design of “moments of impact” [3] by gathering the right people, creating the right conditions, and actively facilitating strategic conversations that build understanding, generate options, and support decision-making. The consultant’s ability to “read the room” and take action on the fly to manage conflict is crucial. The outputs of these sessions/workshops are solutions that are relevant to the customer and an implementation plan that has buy-in from all stakeholders.
Underlying all these jobs is the human capacity for iterative learning. This high-level ability to selectively collect and structure data, reformat it into meaningful knowledge, and apply it to impact our environment is what differentiates us from all other living beings on this planet. This is what we call intelligence; and as our understanding of where these abilities come from improve, opportunities arise for us to artificially augment (or replace) the “jobs” performed by intelligence — as we have done with simpler biological processes (e.g. IVF for reproduction, prosthetics for mobility, and CRISPR/Cas9 for gene mutation).
Artificial Intelligence as a Tool
Until recently, the concept of artificial intelligence (AI) was mostly limited to the science fiction scene with books and movies like “2001: A Space Odyssey”, “The Terminator”, and “The Matrix”. The simple fact was that we lacked sufficient knowledge (mathematical algorithms) and resources (computing power, datasets) to realistically (and cost effectively) make AI useful. However, in the last few years, all of that has changed. Research in Canada [4] has advanced the academic field by leaps and bounds. Moore’s Law and the development of the GPU has reduced the time and cost of processing, and the exponential increase of users on the Internet, and sensors in mobile devices and global infrastructure provides us with more data than we know what to do with. As one IBM report recently stated: “90 percent of the data in the world today has been created in the last two years alone” [5].
Today, artificial intelligence — particularly machine learning (ML) — is one of the hottest topics in industry, politics, and even society as a whole. Breakthrough applications like voice recognition, natural language processing, and robotics give ML the potential to disrupt businesses from healthcare and manufacturing, to transportation and financial services. While the concepts behind ML are daunting, in the coming days it will ultimately prove to be cheaper, more efficient, and potentially more impartial in its actions than human beings [6]. Companies that don’t build up their understanding and capacity to use these technologies will find themselves going the way of Blockbuster, Kodak, and Barnes & Noble.
Machine learning describes a computer’s ability to keep improving its performance without humans having to explain exactly how to accomplish all the tasks it’s given [7]. The system starts with a basic model of how input data should be interpreted and output presented, then is “trained” to:
Make better predictions in the future using known answers (supervised learning) or context & actions (reinforcement learning)
Discover inherent groupings — i.e. clustering, or descriptive rules i.e. association (unsupervised learning)
In recent years, a specific implementation of machine learning, called Deep Learning (DL), has risen to the forefront with its ability to reach near-human level accuracy. It’s highly likely that you’re experiencing some application of DL in your everyday life. Whether it be Facebook’s image auto-tagging, Netflix’s movie recommendations, or interacting with customer service chatbots, AI applications like Deep Learning are finding more and more ways to augment or replace human intervention.
How ML can Augment the Jobs of Design Consultancies
Traditional management consulting firms invest heavily in human capital to perform literature reviews and quantitative data analysis. These “consultants gather, clean, process, and interpret data from disparate parts of organizations. They are very good at this, but AI is even better.” [8] With the improvement of machine learning models and an exponentially growing digitized knowledge base, the role of the traditional consultants/analysts is changing. Skills like emotional intelligence and inductive reasoning, hallmarks of a design consultant are becoming more essential in the age of AI [9]. In fact, the iterative nature of the Jobs of Design Consultancies — Research, Translation, and Facilitation — are mirrored in the way Deep Learning functions for an organization. Data is collected and analyzed, outputs are presented and validated, and ideally, all relevant perspectives and facts are taken into account. Several leading applications of DL in industry can easily find their way into augmenting the design consultant’s work, and even creating new possibilities for what they can do. Besides the obvious automation tasks currently available such as audio transcription and language conversion, other more impactful applications are just around the corner:
RESEARCH: COMPUTER VISION & IMAGE SEARCH
You’re doing research to understand habits around personal hygiene among new immigrants. You’ve had the opportunity to visit a few houses to interview them and observe their personal routines. Many others declined the visit due to privacy concerns, but did offer to send photos of their bathrooms. During some of your in-house visits, you notice something particular: while most of their products are brands that originate from their home countries, their toothpaste and mouthwash were local brands. What if you wanted to find out if the other subjects from different countries had the same purchasing behaviour? What if you wanted to compare other aspects of their bathrooms like cleanliness or storage locations for various items? For a small data set, you could go through their photos one at a time, but what if you were time constrained, or the sample included thousands of participants?
Machine learning (ML) systems already have the ability to identify objects and classify scenes in photographs/videos (computer vision). Couple that with image search (e.g. Google Photos) and researchers will be able to compare physical locations and objects with unprecedented levels of granularity.
TRANSLATION: BEHAVIOUR PREDICTION
You’re trying to make some connection between what you’ve learned from the research and how you can translate those insights into business opportunities. From your interviews, you recognize a pattern of behaviour. Whenever a customer is about to go out for local cuisine, they are more likely to take extra time and effort into looking nice and smelling clean. You want to understand what might be causing this behaviour, and have a couple of assumptions: perhaps they only eat local when it’s a special occasion. Perhaps they feel more comfortable being themselves when they eat at their own ethnic restaurants. Perhaps they feel like there is a higher expectation of cleanliness. What if you could validate these assumptions without confronting these hidden biases?
ML systems already have the ability to predict future behaviours (personal digital assistants) and provide recommendations, like checking your scheduled appointments and letting you know when you should leave to avoid traffic, and even helping to call an Uber. A specific machine learning model could take scheduling data from a calendar and location/movement data from a phone or wearable (e.g. Apple Watch, Fitbit, etc.) to predict the type and duration of activity dedicated to personal hygiene.
FACILITATION: EMOTION DETECTION
You’re hosting a meeting with a diverse group of stakeholders from your client organization, their partners, and customers. Each has full understanding of the problem to solve and have been gathered to ideate on possible solutions. Everyone is split into mixed groups so that each stakeholder perspective can be present. As is usually the case, there are more groups than facilitators, and you’re finding it difficult to know when to rotate to help groups that are stuck or have a conflict that needs to be resolved. Your experience allows you to do so quite effectively, but part of your attention is always allocated to sensing the rest of the room, even when you’re working with one specific group. What if there was a way to monitor the other groups and be notified when you’re needed, without keep your eyes and ears on the rest of the room?
ML’s ability to read human faces and interpret emotions can help facilitators better manage the dynamics of a meeting or workshop. In the near future, ML systems could use video and audio captured in the room to sense when a table is in need of facilitation. A signal could be sent to your phone or watch to vibrate as a notification. Not only could the system predict conflict BEFORE it occurs, the understanding you have that you will be signalled when needed, allows you to be fully in the present.
It may feel like these technologies are still far away, but it’s right around the corner. Affective, a company that grew out of MIT’s Media Lab, is able to analyze facial expressions and emotions using a standard webcam. While, Avigilon, a Vancouver-based company, has developed video analytics technology to help you classify, sort, and search specified people or objects from hours of footage. It won’t be long before these technologies are extended to real-time support systems.
AI needs Humans, Humans need AI
A basic rule of thumb when thinking about the potential of AI is that the highest impact applications will be discovered, not by our thinking alone, but through collaboration between human and machine. Deep Learning systems have the ability to do unimaginable things, for example, Google’s Machine Language Translation system created its own new language to make itself better at its job [10]. In human-centred design, the ability to create new ideas and insights comes from the merging of diverse thoughts and perspectives into a pool that is then made sense of. AI will just be another cog in that wheel.
The ability for an organization to succeed in the not-too-distant future lies in its ability to seamlessly integrate the strengths of its human and machine capital. For large corporations the culture change required to implement AI will be daunting [11]. I spoke earlier of the iterative nature of machine learning (ML) systems — the need to constantly train the model with new data and update the underlying algorithms is necessary. In this way, ML approaches are inherently probabilistic, rather than deterministic. Leaders accustomed to making decisions one-time and watching everything unfold from there will find it challenging to keep up. The ability to be adaptable, have a mindset of iterative learning, collect and share data, and hire the right talent is the key to success.
There are still limits to what AI can do today, but improvements are coming fast and furious. Businesses must start building the people and processes to support an integrated HCD+AI approach to solving their customer’s challenges, or risk becoming obsolete. The future is Human-Centred (AI Powered) Design.
I’d like to hear from you!
What did you find most interesting?
What topics or explanations are missing?
What would you like to learn more about?
Please post a comment to help me provide better content.
References:
Bellis, Rich (2016) “The Low-Tech Problem-Solving Secret That’s Staring Your Company In The Face.” https://www.fastcompany.com/3065178/the-low-tech-problem-solving-secret-thats-staring-your-
Anderson, Ken (2014) “Ethnographic Research: A Key to Strategy”. https://hbr.org/2009/03/ethnographic-research-a-key-to-strategy
Townsend, Tess (2017) “Why Google’s newest AI is setting up in Canada” https://www.recode.net/2017/7/5/15923448/google-deepmind-new-ai-team-canada-university-alberta
Watson Marketing (2017) “The Key Marketing Trends for 2017” https://www-01.ibm.com/common/ssi/cgi-bin/ssialias?htmlfid=WRL12345USEN
Kolbjornsrud, Vegard; Thomas, Robert J.; Amico, Richard; (2017) “How Artificial Intelligence Will Redefine Management” https://hbr.org/2016/11/how-artificial-intelligence-will-redefine-management
Brynjofsson, Erik; Mcafee, Andrew (2017) “The Business of Artificial Intelligence” https://hbr.org/cover-story/2017/07/the-business-of-artificial-intelligence
Libert, Barry; Beck, Megan (2017) “AI May Soon Replace Even the Most Elite Consultants” https://hbr.org/2017/07/ai-may-soon-replace-even-the-most-elite-consultants
Hess, Ed (2017) “In the AI Age” https://hbr.org/2017/06/in-the-ai-age-being-smart-will-mean-something-completely-different
Coldewey, Devin (2016) “Google’s AI translation tool seems to have invented its own secret internal language” https://techcrunch.com/2016/11/22/googles-ai-translation-tool-seems-to-have-invented-its-own-secret-internal-language/
Gerbert, Philipp; Reeves, Martin; Steinhäuser, Sebastian; Ruwolt, Patrick (2017) “Reshaping Business with Artificial Intelligence” https://www.bcg.com/en-ca/publications/2017/strategy-technology-digital-is-your-business-ready-artificial-intelligence.aspx