By Phillip Fernberg, ASLA

Should designers care about artificial intelligence (AI) or machine learning (ML)? There is no question the technology is adding texture to the current zeitgeist. Never could I have imagined seeing a blockbuster hit where Ryan Reynolds emerges as a conscious non-player character in a video game and a flop where Melissa McCarthy negotiates humanity’s future with a James Corden-powered superintelligence within a year of each other. But does learning AI and ML’s ins and outs really matter for the creative professions and our nebulous, invaluable way of operating?

Helen Armstrong, a professor of graphic design at NC State, thinks so. In fact, for her it is imperative. “[AI] is everywhere and has already transformed our profession,” the preface to her new book reads. “To be honest, it’s going to steamroll right over us unless we jump aboard and start pulling the levers and steering the train in a human, ethical, and intentional direction.” The book is Big Data. Big Design. Why Designers Should Care about Artificial Intelligence and its gospel is a primer for designers of all cuts — landscape, graphic, industrial, or otherwise — to get oriented to a brave new world of human-machine relations.

When I say gospel, I do not mean it ironically. Armstrong’s prose is tinged with the passion of an evangelist trying to open our eyes to the great and terrible possibilities of AI-driven design practice. A book of this nature is sorely needed. As Brent Chamberlain and I argued last year in a Landscape Architecture Magazine article, the built environment professions are in the midst of an unprecedented technological transformation that is so overwhelmingly expansive yet so subtle it can be easy to ignore — even if for the mere sake of mental and emotional preservation.

We landscape architects need some particular stirring in this regard. The complexity and timescale of our working medium combined with a mostly healthy skepticism towards new technology for new technology’s sake can sometimes make it seem like the profession is perpetually playing catch-up. Big Data. Big Design. offers the catch-up without condescension, taking the generalist view that every design discipline needs to understand machine learning better regardless of pre-existing technical prowess.

The book’s structure is straightforward, with four main sections sandwiched by a preface and conclusion. The scale of discussion in these sections oscillates between broad definitions of what exactly AI and ML are (Armstrong uses the terms AI and ML interchangeably) and more specific examples of how they are used in design practice.

The parishioner’s tone of the first three chapters then turns more technical in the fourth as the author delves more into the weeds of ML, specifically the differences between its three main approaches: supervised learning, unsupervised learning, and reinforcement learning. If I were to use a crude analogy to sum up the book’s conceptual sequence, I would say it follows Simon Sinek’s golden circle model: it starts with why designers should care about ML, elaborates how designers might use it, and culminates in what such a process might mean for society.

Nearly anyone who lives in the modern world produces data, often on the order of terabytes per day. We text our friends, stream videos, use fitness apps, ask Siri about the weather while we look out the window, walk by CCTV cameras, the list goes on. Most of these data are unstructured, i.e. not organized in any clear order. Machine learning provides a way for computers to glean meaning from this lack of structure.

As Armstrong puts it, “even now as you read, computers sift and categorize your data trails—both unstructured and structured — plunging deeper into who you are and what makes you tick.” How does it do this? The short answer is algorithms, statistical analysis, and prediction. Not sure what any of those words mean? Fear not! The book is riddled with basic definitions in the margins, inset snappy diagrams, and clear infographics that will bring even the most tech-averse designer up to snuff. For some, these visual aids may seem trite, but to me they were integral.

As a researcher dedicated to demystifying emerging technology for landscape architects, I believe it is vital we get designers of all demographics and digital abilities to a shared understanding of what AI is so we can all better facilitate its continued permeation into practice. Big Data. Big Design. does this is in spades.

The book’s real strength lies in the compilation of concrete examples from ML-assisted design practice. Armstrong assembles a fantastically broad collection of work exploring this new era of human-machine design that gives support to her claim that “our interactions with machines are shifting from ‘transactional’ to ‘relational’,” and that with that transition comes a fundamentally new way of seeing design.

The reader is introduced to a vibrant, emerging ecology of human-machine design partnerships, envisaging at once all the good that can be accomplished for humanity when those partnerships are well thought out and all the ill that can come when they are not. There are in-depth interviews with human-computer interaction experts like John Zimmerman and descriptions of visionary creative work like that of Tellart and Toyota’s emotionally intelligent concept cars.

Another example: Superflux’s Mitigation of Shock installations portrays a post-humanist model for adapting to climate change.

And Kate Crawford and Vladan Joler offer a mini-essay on AI ethics.

Besides more minor complaints about lumping ML and AI together as one term, which is not my favorite to see as a technophile but tolerable, or a tendency to occasionally slide into less-than-nuanced conjecture about the implications of technology for society, the most glaring fault a landscape architect will likely see while reading is the omission of ML-driven design being produced in our discipline.

While certainly sparser than that of graphic arts, industrial design, or even architecture, human-machine design work does exist in landscape architecture. Landscape architects are using ML to iterate streetscape designs, explore novel approaches to coastal terraforming, and generate high-level urban design concepts, to name a few things. An author professing to speak to all of us ought to do some due diligence on that, and if she did, at least mention it — especially when she resides in a school that includes landscape architects and is theoretically aware of our impact as a design discipline.

Despite this criticism, it is hard to overemphasize the importance and utility of a book like Big Data. Big Design., which takes an overwhelmingly complex and technical subject and translates it into accessible language for designers of any discipline so that we can better understand how it affects us. The increasing spread of AI into every industry means that those who program AI systems in many ways design the societal outcomes those systems produce, even when said systems become completely autonomous. I agree with Armstrong when she writes “we human designers must be there to frame the right problems — the problems that will move us toward future points that truly benefit humanity.”

Phillip Fernberg, ASLA, is a writer, designer, and PhD Candidate in Landscape Architecture and Environmental Planning at Utah State University, whose work focuses on technology, culture, and design of the built environment.

By Chris S. Sherwin, ASLA

The rise of consumer-friendly unmanned aerial vehicles (UAVs) have enabled landscape architecture firms to take advantage of new technologies and digital processing that have previously been out of reach. The growing availability of drones for designers makes the new book Drone Technology in Architecture, Engineering, and Construction by two landscape architects — Daniel Tal, FASLA, and Jon Altschuld, ASLA — a timely addition to any firm’s library.

The book enables a landscape architecture firm with little drone experience to understand step by how to work a drone and the opportunities and obstacles with owning and operating one. Tal and Altschuld have transformed a complex and obscure subject into accessible content. The book is meant to be read in a sequential manner: each chapter builds upon the other and offers resources for further study.

Tal and Altschuld begin the book by explaining the rapid transformation of UAVs from a military tool to hobby radio-controlled aircraft. Consumer-level drones enable design professionals and clients to better understand the environment from the bird’s eye view. Drones offer unlimited potential for imagery and photogrammetry within a 3D modeling environment and are an important tool for communicating how a proposed design can impact sites.

The risks and shortfalls of drone technology are outlined through an exploration of the difference between good looking data and good data. With new low-cost drones, site scanning often involves just pressing a couple of buttons on an app and setting the site parameters. A user can produce a very detailed and good looking data set without understanding how these technologies function. This can be a hindrance to producing good data.

A new drone user needs to understand that producing a precise and correct data set involves a deeper understanding of drone technology, surveying, and photogrammetry processing. An accurate model requires knowledge of how to set ground control points, provide specific camera angles in relation to the site being surveyed, and set a steady and continuous ground sampling distance from the site as the drone surveys the property. Fortunately, all of these areas are thoroughly covered by the authors.

The book further explores how to effectively use drone data visualization as a project life-cycle tool. Chapters explain the costs necessary for staff training, examinations, and certifications; purchasing an appropriate drone; the software needed for flight, video and photo recording, photogrammetry; and insurance prior to your first flight.

Documentation, permissions, and licensing are also explored in-depth. The authors review the legal rules pertaining to consumer and commercial drone operations. Most importantly, they cover the Small Unmanned Aircraft Rule, which is known as 14 CFR Part 107, or in industry parlance the Part 107 exam. This exam is a crucial step in becoming a full-fledge drone operator in the design professions. Another vital step before becoming a commercial operator is understanding the permissions required through the Low Altitude Authorization and Notification Capability (LAANC) system, which allows for a relatively quick and seamless on-line process to gain permission before flying on site.

Proper documentation needs to be carried at all times. These documents include insurance certificates, your remote pilot license (Part 107), UAS registration, and a log of your flight operations. This documentation is key should something go wrong during your flight, and it is always a good rule of the thumb to be prepared for the worst case scenario.

Through the book, readers will gain an understanding of best flying practices, which many will gain with experience over time, and the importance of an appropriate safe flying mindset, such as developing a procedure for flight operations prior to, during, and at the conclusion of the flight. Situational awareness is key as well as proper flight procedures during manual and automated flying.

Tal and Altschuld explore basic drone photography, which many drone operators will use in the beginning of their flying career since it is relatively straightforward. Plus, the photographic results can be easily adjusted in commonly available Adobe Acrobat or Bluebeam software. Photo match overlays of the proposed project can easily be inserted into the existing site using Photoshop. 3D model photo overlays can also include integrating 3D modeling software models of the proposed design into the existing drone photography. This is a great tool for project visualization and client presentation meetings in which drone visualization images help communicate complex ideas easily and efficiently.

Chapters on photogrammetry and 3D modeling are the most crucial, because these tasks will ultimately provide the most comprehensive data needed for advancing complex design projects. The catch is that photogrammetry, although simple in concept, is a relatively complex process that requires time and patience. The first-time drone user needs to make sure that the data acquired is consistent and free of bugs; otherwise, the end result – the 3D mesh point cloud — will be rife with topographical errors, and the resulting model will be inaccurate and not suitable for using as a 3D Mesh in Sketchup, Autodesk Revit, or ReCap.

Precision and patience are necessary to produce a highly refined and detailed 3D point cloud. The author explain the steps required for an accurate on-site scan, such as establishing ground control points, proper flight, image collection, and weather planning. These methods require further study and a trial period of trouble shooting. The drone user will most likely produce several versions of a photogrammetric model before arriving at a finished product.

Drone Technology in Architecture, Engineering, and Construction is well-written and concise enough that the reader will not get bogged down with the details but still be engaged throughout the process. It is perfect for landscape architecture firms seeking to purchase a drone but are unsure of next steps.

Chris S. Sherwin, ASLA, RLA, is managing director and drone mapping expert at CSS D/S LLP in Oakland, California and Ulaanbaatar, Mongolia.