Earlier this year Randy Deutsch gave a talk at the DFC Leadership Summit on Design Innovation & Technology 2018 in La Jolla, CA, “Toward an Augmented Architect: Learning from Machine Learning, Embracing and Capitalizing on AI.” The heart of the talk, on which this essay is based, were recommendations to firm leaders for making the most of our industry’s technology transformation.
Designing in a Moleskine or on yellow trace for many is still de rigueur. For others, designing has changed due to the introduction of new computational tools that leverage data, algorithms, and the cloud. But the technology is not what’s driving this change; it is risk. On project teams that are taking a design from a state of uncertainty and ambiguity to one of certainty and clarity, architects may be comfortable with ambiguity. However, owners still require certainty.
Design: data-driven, generative, and predictive
To address the need for certainty, many of the activities designers do today are being transformed into data, and many tasks of the design process are being automated. With data-driven design, we design by manipulating data, not form. With generative design, we design by leveraging algorithms and parametric modeling within predetermined constraints. With predictive design tools—like Autofill—we design anticipating our next move.
Today a growing vocal minority—superusers—speak of scripts and algorithms as integral to their design process. They use visual programming tools to automate and complete work in hours that might otherwise take days. They create cloud-enabled data visualizations as a real-time by-product of the design act, where data not only informs their intuition, but improves it. They design with adaptive parametric smart elements that use rules to govern what the user can do, so when something moves or changes everything moves or changes with it. Indeed, much of our design has already been outsourced—not overseas, or even to people, but to software. Think AI is too long of a time horizon to reasonably address? Just as we can take photographs with apps that will correct and edit them before we’ve even taken them, AI-enabled design can be informed and improved in predesign by predictive post-occupancy evaluations that take place before the project is even designed.
Eight recommendations
AI is fast approaching—coming to your home, car, and office—and it’s coming too fast to fight off or control. Why wait for a fire you cannot put out when there are things you can be doing today to prepare for the inevitable conflagration of AI? Think of these recommendations as your personal fire truck protecting your firm from, and preparing you and your career for, the impending AI inferno.
1. Be concerned and vigilant but don’t be fearful. The concern isn’t that “the robots will take over.” Warranted or not, the concern for emerging professionals in particular is that they want to develop entry-level skills. They’re afraid that, by AI doing this “drudgery work” for them, the de-skilling of practice that frees them “to work on design and be creative” instead keeps them from developing into well-rounded professionals in their work experience flow. We owe it to ourselves, to our clients, and especially those we serve, to be aware and mindful, but not anxious, panicked, or alarmed. As leaders, we need to talk about AI not in terms of survival, but instead, how we’re going to leverage, exploit, embrace, and capitalize on the changes.
2. Design buildings but also processes and algorithms. UNStudio’s Ben van Berkel predicts, “In the future all architecture practices will become arch tech firms, but for now we have to pave the way to make this expansion of our knowledge and expertise possible.” Professional design service firms today must also be software practices and technology firms, creating apps and other digital tools—even if it’s not in their DNA. One reason, beyond increased productivity, is employee retention. A/E/C is undergoing a brain drain, where architecture firms compete for the best people not only with other architecture firms, but also with software developers and startups. In lieu of traditional practice, recent graduates are going into arguably more alluring, lucrative startups and quicker-paced software development companies. How do we keep talented, smart people interested in a field where projects can take 3–6 years or longer to complete?
3. Redefine optimization. We make a mistake when we define optimization as infinite reduction. Over-optimization leads us to over-analysis. Machines, learning or otherwise, are not going to understand the entire context and meaning behind what they’re doing. We sell ourselves short when we just focus on analysis. Architects are not going to out-compute computers. One responsibility of firm leaders is to redefine optimization from something that only means reduction to something that enhances what we produce. Leaders can focus on addressing complex, intractable, wicked problems; tapping multiple minds on multidisciplinary teams; creative acts that create meaning; using common sense—which computers seem to ignore when spitting out optimal solutions; looking at qualitative variables, including all criteria, then weigh, combine, and synthesize. Play to your strengths and multiple intelligences; bolster them and leave the quantitative for the quants. If you don’t like the decision the machine manufactured, then override it. Human override is where you still (for now) have the final say concerning the design. Don’t rely on the most efficient outcome; instead, define what is optimal for you and your firm.
4. Identify opportunities for automation; don’t just focus on what cannot be automated. Why automate? Automation turns a two- to three-day assignment into a twenty-minute step. Firms need to automate what they repetitively generate manually, then look through their standard delivery processes to see how much more they can automate. Today we can automate the manual calculation parts of design, addressing ergonomic standards and legal code requirements for life safety. So automate, but as Ian Keough, the founder of Dynamo suggests: Architecture isn’t what’s left over after everything’s been automated.
5. Collaborate with technology. Mentioning collaborate with technology often gets eye-rolls, such as this reply-tweet: “Myself along with hammer and table saw built a house. I also need to give a shoutout to lumber, we nailed it.” That said, the best chess player in the world recently was neither a chess master nor a computer, but a couple of teenagers with a laptop. So, how will you prepare your firm to address projects in the future? What skills will be valued in this new era? How can architects stay agile and steer their careers through this time of unprecedented change? It’s likely to involve working side-by-side with machines. Leveraging man-machine collaboration—letting machines and architects do what they each do best—will achieve new levels of productivity for both.
6. Help others transition to AI. Leading firms in A/E/C are now redesigning their work processes to be faster and more adaptable. To achieve these gains, automating current work processes and tasks won’t be enough. To increase efficiencies, architects will need to transition to working with AI, which will require us to move beyond linear, sequential, and repeatable processes toward employing adaptive project teams (APT) that pair architects with real-time, data-driven AI systems. Leaders must lay the groundwork, identifying differences between their traditional processes and new thinking, and encourage others—especially those not trained as technologists—to get acquainted with the technology.
7. Augment – not replace – architects. Teach machines to think like us? We’re still working on teaching us to think like us. Besides, machines will think any way they want to. Do we even want machines to think like us? Instead, let machines do what they do best. We have hybrid cars and hybrid buildings; now we are starting to look at hybrid individuals, including augmented and virtual architects and operators. Architects are going to become augmented architects in the labor force. Architects who view themselves as mixologists see this as an incredibly creative opportunity to mix, play with, and converge technologies for better results. They see architecture as more than a series of interconnected tasks, and architects as more than the sum of their parts. Don’t ask: Will architects be better off if the machine decides this? Instead, ask: Will our clients and users be better off if we decide, or if the machine decides? Will our clients be better off if the machine decides this in conjunction, or collaboration, with an architect?
8. Think like machines (and everybody else for that matter). It’s not just the output of machines that we’re benefiting from, it’s also the way machines learn and think and adjust on their own. We have a lot to learn from them. Thinking like others on project teams makes you a more effective communicator and empathetic team member. By understanding what is important to others—including machines—we can more effectively design and shape our message.
Design will continue to change in this time of interdisciplinary collaboration, when project phases are merging, disciplines are blurring, roles are blending, and tools are converging; when architects are moving into means and methods, builders are providing design services, and design, fabrication, and construction are becoming increasingly indistinguishable. One can soon imagine design no longer serving as a standalone phase, and design technologists become (once again) just technologists. Becoming augmented architects is our best hope as a profession and industry.
As I ask this question: As the next generation of practitioners with next generation solutions, what role do you want to play?
Designing in a Moleskine or on yellow trace for many is still de rigueur. For others, designing has changed due to the introduction of new computational tools that leverage data, algorithms, and the cloud. But the technology is not what’s driving this change; it is risk. On project teams that are taking a design from a state of uncertainty and ambiguity to one of certainty and clarity, architects may be comfortable with ambiguity. However, owners still require certainty.
Design: data-driven, generative, and predictive
To address the need for certainty, many of the activities designers do today are being transformed into data, and many tasks of the design process are being automated. With data-driven design, we design by manipulating data, not form. With generative design, we design by leveraging algorithms and parametric modeling within predetermined constraints. With predictive design tools—like Autofill—we design anticipating our next move.
Today a growing vocal minority—superusers—speak of scripts and algorithms as integral to their design process. They use visual programming tools to automate and complete work in hours that might otherwise take days. They create cloud-enabled data visualizations as a real-time by-product of the design act, where data not only informs their intuition, but improves it. They design with adaptive parametric smart elements that use rules to govern what the user can do, so when something moves or changes everything moves or changes with it. Indeed, much of our design has already been outsourced—not overseas, or even to people, but to software. Think AI is too long of a time horizon to reasonably address? Just as we can take photographs with apps that will correct and edit them before we’ve even taken them, AI-enabled design can be informed and improved in predesign by predictive post-occupancy evaluations that take place before the project is even designed.
Eight recommendations
AI is fast approaching—coming to your home, car, and office—and it’s coming too fast to fight off or control. Why wait for a fire you cannot put out when there are things you can be doing today to prepare for the inevitable conflagration of AI? Think of these recommendations as your personal fire truck protecting your firm from, and preparing you and your career for, the impending AI inferno.
1. Be concerned and vigilant but don’t be fearful. The concern isn’t that “the robots will take over.” Warranted or not, the concern for emerging professionals in particular is that they want to develop entry-level skills. They’re afraid that, by AI doing this “drudgery work” for them, the de-skilling of practice that frees them “to work on design and be creative” instead keeps them from developing into well-rounded professionals in their work experience flow. We owe it to ourselves, to our clients, and especially those we serve, to be aware and mindful, but not anxious, panicked, or alarmed. As leaders, we need to talk about AI not in terms of survival, but instead, how we’re going to leverage, exploit, embrace, and capitalize on the changes.
2. Design buildings but also processes and algorithms. UNStudio’s Ben van Berkel predicts, “In the future all architecture practices will become arch tech firms, but for now we have to pave the way to make this expansion of our knowledge and expertise possible.” Professional design service firms today must also be software practices and technology firms, creating apps and other digital tools—even if it’s not in their DNA. One reason, beyond increased productivity, is employee retention. A/E/C is undergoing a brain drain, where architecture firms compete for the best people not only with other architecture firms, but also with software developers and startups. In lieu of traditional practice, recent graduates are going into arguably more alluring, lucrative startups and quicker-paced software development companies. How do we keep talented, smart people interested in a field where projects can take 3–6 years or longer to complete?
3. Redefine optimization. We make a mistake when we define optimization as infinite reduction. Over-optimization leads us to over-analysis. Machines, learning or otherwise, are not going to understand the entire context and meaning behind what they’re doing. We sell ourselves short when we just focus on analysis. Architects are not going to out-compute computers. One responsibility of firm leaders is to redefine optimization from something that only means reduction to something that enhances what we produce. Leaders can focus on addressing complex, intractable, wicked problems; tapping multiple minds on multidisciplinary teams; creative acts that create meaning; using common sense—which computers seem to ignore when spitting out optimal solutions; looking at qualitative variables, including all criteria, then weigh, combine, and synthesize. Play to your strengths and multiple intelligences; bolster them and leave the quantitative for the quants. If you don’t like the decision the machine manufactured, then override it. Human override is where you still (for now) have the final say concerning the design. Don’t rely on the most efficient outcome; instead, define what is optimal for you and your firm.
4. Identify opportunities for automation; don’t just focus on what cannot be automated. Why automate? Automation turns a two- to three-day assignment into a twenty-minute step. Firms need to automate what they repetitively generate manually, then look through their standard delivery processes to see how much more they can automate. Today we can automate the manual calculation parts of design, addressing ergonomic standards and legal code requirements for life safety. So automate, but as Ian Keough, the founder of Dynamo suggests: Architecture isn’t what’s left over after everything’s been automated.
5. Collaborate with technology. Mentioning collaborate with technology often gets eye-rolls, such as this reply-tweet: “Myself along with hammer and table saw built a house. I also need to give a shoutout to lumber, we nailed it.” That said, the best chess player in the world recently was neither a chess master nor a computer, but a couple of teenagers with a laptop. So, how will you prepare your firm to address projects in the future? What skills will be valued in this new era? How can architects stay agile and steer their careers through this time of unprecedented change? It’s likely to involve working side-by-side with machines. Leveraging man-machine collaboration—letting machines and architects do what they each do best—will achieve new levels of productivity for both.
6. Help others transition to AI. Leading firms in A/E/C are now redesigning their work processes to be faster and more adaptable. To achieve these gains, automating current work processes and tasks won’t be enough. To increase efficiencies, architects will need to transition to working with AI, which will require us to move beyond linear, sequential, and repeatable processes toward employing adaptive project teams (APT) that pair architects with real-time, data-driven AI systems. Leaders must lay the groundwork, identifying differences between their traditional processes and new thinking, and encourage others—especially those not trained as technologists—to get acquainted with the technology.
7. Augment – not replace – architects. Teach machines to think like us? We’re still working on teaching us to think like us. Besides, machines will think any way they want to. Do we even want machines to think like us? Instead, let machines do what they do best. We have hybrid cars and hybrid buildings; now we are starting to look at hybrid individuals, including augmented and virtual architects and operators. Architects are going to become augmented architects in the labor force. Architects who view themselves as mixologists see this as an incredibly creative opportunity to mix, play with, and converge technologies for better results. They see architecture as more than a series of interconnected tasks, and architects as more than the sum of their parts. Don’t ask: Will architects be better off if the machine decides this? Instead, ask: Will our clients and users be better off if we decide, or if the machine decides? Will our clients be better off if the machine decides this in conjunction, or collaboration, with an architect?
8. Think like machines (and everybody else for that matter). It’s not just the output of machines that we’re benefiting from, it’s also the way machines learn and think and adjust on their own. We have a lot to learn from them. Thinking like others on project teams makes you a more effective communicator and empathetic team member. By understanding what is important to others—including machines—we can more effectively design and shape our message.
Design will continue to change in this time of interdisciplinary collaboration, when project phases are merging, disciplines are blurring, roles are blending, and tools are converging; when architects are moving into means and methods, builders are providing design services, and design, fabrication, and construction are becoming increasingly indistinguishable. One can soon imagine design no longer serving as a standalone phase, and design technologists become (once again) just technologists. Becoming augmented architects is our best hope as a profession and industry.
As I ask this question: As the next generation of practitioners with next generation solutions, what role do you want to play?
Follow on Instagram
No comments:
Post a Comment