Published as Norman, D. (2017). Design, business models, and human-technology teamwork. Research-Technology Management, 60(1), 26-29.
I was at Apple's Advanced Technology Group in the 1990s, a most exciting time in technology. In addition to many advances in technology and human-machine interfaces, the Internet was taking hold, web browsers were becoming standard, the first Wi-Fi band was established, along with today's HDTV standards. That was a time of great changes--but today's changes are even more exciting.
Consider the rise of intelligent machines, or the growth of cloud services coupled with the prevalence of computer chips, sensors, and telecommunications capability in almost everything: the Internet of Things. Technology is undergoing rapid change, which means that business models must change as well. Think of the new world as EaaS: Everything as a Service.
Advances in artificial intelligence and machine learning coupled with cheap sensors, ubiquitous telecommunications, and ever more powerful computer chips have come together to enable unprecedented automation of tasks long thought undoable by machine. Machines are increasingly taking over tasks once done by people, to the extent that theorists are beginning to construct a framework for a "post-work" economy. More and more jobs require people to work closely with machines, handing off tasks from person to machine and then back again, supervising the work of machines, and sometimes being supervised by machines.
As the technologies of automation and artificial intelligence continue to develop, the concerns and challenges associated with them will become the province of R&D in every industry. Sometimes new products will offer increasing levels of automation. Sometimes the way that we design for manufacturing must be changed to allow for automated "mass customization" in production. All these factors will require fundamental change. Business models must change. Product designers and engineers to work differently.
We now face huge challenges and opportunities. How do we mitigate the challenges and take advantage of the opportunities? What happens to conventional business and to conventional manufacturing? I would argue that we need to think differently about technology. We need to think about automation that enhances people versus automation that replaces people. We need to think less about the design of human-machine interfaces (HMI) and more about the design of human-machine teamwork. And we need to continue to expand the concept and tools of design, from design as specification, to design for use, to design that includes the new business models.
Automation That Replaces People versus Automation That Enhances People
Some automation replaces the tasks that people do. Other automation enhances the work of people, making people more capable and effective. When is each relevant? One rule commonly discussed in robotics is that automation should replace people for tasks that fall into one or more of the three Ds: dirty, dull, and dangerous. But for other tasks, those that are clean, exciting, creative, and safe, automation should be used to augment human abilities.
Routine skills, whether manual or mental, can be automated (and in many cases have already been), but this kind of automation does not lessen the need for highly qualified people. Indeed, it enhances their abilities, making them even more valuable. Let me give a trivial example.
The development of calculators (from arithmetic through calculus) and computer systems did not eliminate the need for people with mathematical training. What it did do was eliminate the kinds of clerical errors even great mathematicians make. The machines and programs do the math; the mathematician concentrates on figuring out what problems should be solved. The human mind is far more powerful when coupled with the smart tool. The combination is far superior to either one alone. This observation, which is the core argument of my book Things That Make Us Smart, has two implications:
- We need to redefine the jobs people do to take advantage of the synergies afforded by the combination of people plus tools.
- When we design new intelligent tools, we should think of them as collaborators, not as replacements.
For those working at the forefront of these technologies, there is a tendency, often subconscious, to be so enamored of the challenge of automating activities and tasks that the human element is ignored. Or, perhaps worse, the human element is used as the escape hatch: automate whatever can be automated and leave the rest to people.
Too often, we design technology to do whatever it can, leaving the rest to people. This means that people have to fill in the gaps in machine performance, requiring that people behave on the machine's terms--working at the machine's level of precision and accuracy, at the time required by the machine. And they must continually monitor the activity so they can take over when necessary. These are all things people are bad at. The result is that we ask people to do the things they are bad at, and when they turn out to be bad, we blame them: human error, we say. Self-fulfilling prophecy, I say. The problem is design error, born of a misconception of the proper role of technology.
The key is to start with human capabilities, designing the technology to take over the parts of a task that people are bad at. Let people decide upon high-level goals and constraints. Rely on machines to supplement the truly astonishing capabilities of humans.
Technology and human intelligence have different characteristics, different strengths, and different ways of working. Machines process information very quickly, never get bored, and reliably do the things they are designed to do. People excel at tasks requiring the exercise of creativity, a response to unexpected situations, or general attentiveness to the entire surrounding environment. A truly powerful automation approach takes these different strengths into account to create a superior, collaborative system.
An Example of Human-Technology Teamwork
Today's design tools require the designer to specify almost all of the details of the desired product. Modern design tools can compute stress and strain, heat transmission, and other parameters of physical models and electronic characteristics of circuits. But the designer must still do the low-level drawing.
Consider instead a situation where the designer specifies the high-level constraints and goals--size, weight, volume, strength, heat dissipation, and other characteristics, perhaps ergonomic requirements or task information to help configure controls and displays--then let the design tool suggests solutions. This would be a dramatically different way to design than what we do today. The designers would specify the high-level parameters--goals and constraints--and the tool would generate potential solutions. Then, as solutions are generated, the designer would assess them, encouraging some directions and discouraging others. If the suggested solutions did not satisfy the designer, the system could be given modified goals and constraints, or perhaps more restrictions, to force the solutions into a desirable set of alternatives.
These design tools would work by using machine learning algorithms (genetic algorithms, neural networks, Bayesian weights) to modify the generation process. The machine and the designer would be true collaborators, each doing a very different part of the task, but together closing in on a desirable solution, perhaps one never imagined by the designer. Is this pure fantasy? No. Autodesk, for example, has been experimenting with systems just like this. Their experimental system Dreamcatcher uses a procedure similar to my description; they call the process "generative design":
Generative design is software technology that lets you create highly optimized designs that meet predetermined goals and constraints. Using shape synthesis algorithms and multiphysics performance analysis in the cloud, the software generates thousands of design options--many that you'd never think of on your own--from a single idea.
Once again, collaboration.
The same kind of approach will emerge in other fields. Consider medical diagnosis, which with the arrival of precision medicine and personalized information about the patient's genomes, microbiomes, environment, and activities, can no longer be fully analyzed by unaided physicians. Computer systems can use machine learning algorithms similar to the ones for machine design, but with deep knowledge of the relevant medical literature. Numerous systems that do this are in development today, some are in use.
Design, Usability, and Productivity
What does design have to do with all of this?
Accomplishing this synergy between human and machine requires new attention, and a new kind of attention to design. Design is a badly misunderstood term. To the engineer, design is the detailed specification of components or products that will be implemented, whether as a manufactured product, a computer program, or a service. To the public, design means making things look pretty. Indeed, a recent Wall Street Journal article quoted someone as saying "I would rather take a good design than a good function" ("Sleek Italian Kitchens Cross the Pond," Sep. 16, 2016 p. M8). I nominated the quotation as "stupid statement of the week" in a tweet that received huge positive response starting seconds after it was sent. A design with poor function is poor design.
Design, however, is much more than technical specifications or appearance. It encompasses issues of usability and productivity that are becoming ever more critical with the rise of increasingly complex products that require humans to interact with them in ever more complex ways. For many modern designers, design is a way of thinking, of ensuring that we are solving the correct problem and that the design really does match the true, underlying need of its eventual user. Modern design is people-centered, focused on the needs of those who will interact with the product or service. These needs and coupled with the abilities of people are foremost in the minds of good human-centered designers, as they engage in an iterative process of creatively developing and refining solutions.
This focus on people is critical because every product or service sold involves people. Even things thought to be independent of people seldom are: Machinery has to be installed, maintained, repaired, and upgraded. Satellite dishes have to be aimed with precision, usually manually. Spreadsheets and electronic medical records have to be interpreted by people, often people who have limited time and divided attention. Bad design leads to bad performance on these tasks, and often to error. Good design doesn't just stay out of the way; it makes people better at their work.
But too often, designers are allowed in only at the end of the process (to make it look pretty) rather than at the front end (to make sure that it's the correct product; that it's understandable, usable, and desirable; and that it doesn't need frequent service calls or have complex maintenance requirements). These are all aspects of design, and they are all natural results of good design, of the integration of design thinking into every aspect of product planning and development.
Increasingly, the preeminent design issue is design of the business model. Designers will need to move from what deciding how to create our existing products to using design research to determine what the real needs will be for our customers, what new business models will allow nimble, relevant satisfaction of those needs.
Think of Uber: Before Uber, the San Francisco taxi market was around $140 million a year. How much of that do you think Uber took? Answer, according to Uber's CEO: $500 million (Blodget 2015). People are using Uber in ways they never would have considered for a taxi. Business models need to change.
Start by questioning today's business models. Ask, what are the alternatives to what is done today? Follow the startup markets: what new ideas are being generated there. The most powerful ideas will often come from companies that we do not take seriously today. Business models do not normally fall into the responsibility of research directors, but the business model structures all the products and services we provide: it determines the form, shape, and operation of the company. It should be a major component of a research director's concerns.
Conclusion
As automation enters the workplace and our lives, eliminating some tasks (such as driving) and empowering others, and altering the type of products and services we offer and the means by which they are offered. The Wall Street Journal quote about good design versus function that I cited earlier is stupid because good design is good function. A good design provides people with a good conceptual model for what is happening, with feedback and visibility that promotes understanding and a feeling of control. This is especially important when things go wrong. It is relatively easy to design something that works fine as long as everything goes well; but as soon as an unexpected event occurs, some accident or even an input or command error, poor design can lead to confusion, which in turn leads to errors and considerable difficulties and cost. Poor design prevents people from taking full advantage of the power of many devices, or requires extensive training for people to access that power: all of these combine to lower productivity.
I believe that design is so critical to ensuring that the correct business decisions are made that it should be one of the major drivers of company strategy: design, engineering, manufacturing, sales, marketing--all play critical roles. Yet too often, design is only incorporated at the end of the process rather than at the front end. Leaders of research, technology, and product design can help to correct this error and drive competitive advantage for their companies.
Leaders of R&D and product development must also reconceptualize their role to include the development of new business models. This will require widening the lens through which they look at the world, redefining the relationships they must build in the business, staying abreast of developments in fields that may seem far away but that have disruptive potential, and taking risks beyond those of product development.
References
Blodget, H. 2015. Uber CEO reveals mind-boggling new statistic that skeptics will hate. Business Insider, January 19. http://www.businessinsider.com/uber-revenue-san-francisco-2015-1
Norman, D. A. (1993). Things that make us smart. Cambridge, MA: Perseus Publishing. http://jnd.org/books.html#ttmus
Note
1. See http://www.autodesk.com/solutions/product-development-innovation-platform
Bio: Don Norman is both a businessperson--having served as a vice president at Apple and an executive at Hewlett Packard--and an academic who has taught and conducted research at Harvard, University of California, San Diego, Northwestern, KAIST (Korea), and Tongji (Shanghai). He is currently Founder and Director of the Design Lab at the University of California, San Diego, where several decades ago he founded and was first chair of the department of cognitive science. He is cofounder and principal of the Nielsen Norman group, a member of the National Academy of Engineering, an IDEO fellow, and a trustee of IIT's Institute of Design. He serves on several company boards; has honorary degrees from Delft, Padua, and San Marino; and has been awarded the Lifetime Achievement Award from ACM's Computer-Human Interaction Group and the President's Lifetime Achievement Award from the Human Factors and Ergonomics Society. He has published 20 books, including Emotional Design and Design of Everyday Things. He can be found at www.jnd.org. dnorman@ucsd.edu