AI won’t eat design. At least not yet.
by Aric Cheston | December 16, 2016 | More Articles
Nothing has so thoroughly unnerved so many people working in so many industries as the prospect of fully realized artificial intelligence. It forces us to scrutinize our life’s work and ask what is inherently human. What do I do that can’t ultimately be boiled down to a rote process? Are the skills and intuition I have honed over my career actually nothing more than advanced pattern recognition? Why should I, a human being with all my needs and rights and potential for volatility not be replaced by software which requires only electricity and data and occupies no more space in the physical world than this very premonition of doom?
It is easy to imagine a line in history separating the time before artificial minds and what comes after. On the far side of that line the great professions are diminished; doctors, lawyers, financiers are exposed as masters of arcana precariously presiding over an amount of data that no human could in reality synthesize. Meanwhile, machines, with their infinite capacity for tedium and statistical analysis of unfathomable amounts of information, simply automate the skills it took a lifetime to amass.
In the world of professional design, of creativity, of flashes of insight, of happy accidents that lead to previously unimagined solutions, we hope that we are immune. We hold on to the idea that what we do is too organically human to be replicated by inorganic mechanisms. We want to believe that creativity cannot be automated, that it takes a human mind to even imagine an artificial one. The troubling fact is that we view creativity as ineffably human because we don’t fully understand it. We know so little about how the brain creates the mind that it is premature for us to assert that what we do cannot one day be replicated by machines.
Certainly, there is no small debate over when a general artificial intelligence will be achieved, if it’s even possible, and what its effect on human civilization will be if it is. Theories range from the full achievement of human potential to the usurpation of our place on the planet. To ground the conversation, it’s useful to consider our experience of automation to date. It is impossible to deny that technologies for automation have created massive social upheaval and economic displacement. The introduction of robotic assembly lines is an easy example to bolster the argument that machines will inevitably replace people. However, the history of automation is hardly a linear march towards full mechanization. In ‘Our Robots, Ourselves: Robotics and the Myths of Autonomy’, MIT professor David Mindell points out that many technologies intended to replace humans are eventually repurposed into systems that are built around a human operator. Automation technologies find their best application when they are augmenting people, not replacing them. Given the state of artificial intelligence in this present moment, augmentation seems like the most productive model as we consider the near term and tangible implications for design.
While the viability of general AI might be debatable, many of its fundamental building blocks are being built and productized. Machine learning, the general term for algorithms that identify and classify patterns within large amounts of unstructured data, is the most fundamental and the most potent of these. And we will very soon see how machine learning will augment not only what we design but how we design.
Machine learning is already shaping the next generation of design tools. Next year, Autodesk will release Dreamcatcher, a Generative Design tool that will let designers generate potentially thousands of permutations of a design within minutes. The algorithms that power Dreamcatcher have ingested an immense amount of 3D data and have been trained to associate that data with discrete design outcomes. This allows designers to create a landscape of viable solutions much broader than one would otherwise be able to in a reasonable amount of time. Generative tools like Dreamcatcher will free designers to direct more energy towards the higher order thinking required to synthesize the most meaningful design from the strongest permutations. Additionally, important standards and guidelines, like those governing accessibility, can be prebuilt into the tools, saving even more time. While Dreamcatcher is a tool for modeling 3D forms, surely a set of tools for assisting in the design of digital products can’t be far behind.
You don’t have to look far beyond projects like The Next Rembrandt (which seems to exist solely to irritate fine artists) to see that machine learning will not only drive a new generation of creative tools, but will be its own creative medium. Creative AI is a crowd-sourced collection of design and art that explores the space defined by the intersection of human creativity and artificial intelligence. In browsing the dozens of projects, it’s possible to perceive the living dialogue between people and technology — that people will always seek to master the machine and turn it towards unexpected and inspired purpose.
Artificial intelligence will certainly have an effect on how and what we design. Just as certain is the vital role designers will play in shaping the relationship between people, their work, and ever more capable machine intelligence. Systems and tools based on this technology need to be built around people, not over them. This will be our job, to apply the human insights that are fundamental to design practice to the partnership of human and machine intelligence. We can start with our own work.
Over the next several months, we’ll be exploring the intersection of design and existing artificial intelligence technologies through experiments, prototypes and concepts. We believe that this is a critically important topic, so we’ll share what we learn along the way. We’re looking for collaborators with interest and background in the application of artificial intelligence to human-centered design, and we’re happy to facilitate any conversations on the topic. You can reach us at email@example.com
As always, we pick the intellectual pockets of giants:
For a refreshingly un-frothy perspective on automation, ‘Our Robots, Ourselves: Robotics and the Myth of Autonomy’ by David Mindel is highly recommended.
Get a great introduction to the principles of Interactive Machine Learning via Greg Borenstein.
For more thoughts on automation and the design process, take a look at a piece by our former colleague Roberto Veronese.
Benedict Evans has great observations on the current and future state of machine learning. Follow him.
To annoy your MFA friends, send them this.