[+] Juggernaut That Won’t Be Stopped: The Coming of AI

As part of his doctoral thesis on Artificial Intelligence in Design, Gabriele Mirra created an application at the University of Melbourne that would build him a bridge. Or at least, design one. Into the application—let’s call it a machine—he fed the parameters for the structure. It had to be 50m long, with a maximum height of 20m. Mirra wanted two, divergent performance criteria, in this case the overall weight of the bridge and its displacement. “So the idea was to have conflicting, let’s say, performance criteria,” said Dr Alberto Pugnale, a senior lecturer in architectural design, and Mirra’s PhD supervisor. “We were looking at the displacement or the deformation of the bridge against the mass, because if you only consider mass, the bridge is simple and more efficient, but heavy,” he said. “But we wanted the final solution to be a balance between a lightweight bridge, and one that wouldn’t deform so much, no more than 20cm.” Mirra’s modelling was part of a thesis chapter he would write on playfulness. Like a small child might learn by repeatedly building a tower from wooden blocks, until it falls, the machine was also encouraged to learn from “play”. “For the child, the moment that tower falls there is sadness, but they are learning something,” Pugnale says. ▲ A concept design for a museum created using OpenAI’s Dalle-2. Planning and design is likely to be the first to take real advantage of AI. “With these AI models the sketch generated is calculated structurally. And if it’s not sound, the model gets a negative reward. If it is structurally sound, it gets a positive word.” It’s a form of “unsupervised learning,” he says. In other words, the model learns from its mistakes. Mirra successfully defended his thesis last year, and today is undertaking post-doctoral research at the University of Southern Denmark, testing new strategies to train AI models for design applications. And the bridge? Using a form of reinforced learning, the AI model came up with a design. In fact, many different designs. “And that’s the difficult thing for us human designers, to make sense of what they come up with because they are all so different,” Pugnale says. “Conceptually speaking, the designs were all sound,” he says. “But I guess what we wanted to do is to understand how different and creative these solutions were.” ▲ Gabriele Mirra’s AI model created this design for a bridge—a cross between an arched and a suspension structure. It’s at least a couple of years before AI can be left alone to design a bridge, and even then, it will likely be in partnership with humans, crunching data and numbers at lightning-fast speeds. But make no mistake—Artificial Intelligence is coming to an office, a studio or a building site near you. And faster than you think. John Butters, vice president and senior earnings analyst with the business data provider FactSet, monitored the earnings conference calls from all S&amp;P 500 companies in the three months to September 7 last year. Butters says 177 cited the term “Artificial Intelligence” during their earnings calls for the second quarter. And none more famously than Silicon Valley chipmaker Nvidia, which last week reported $22.10 billion in revenue for its fiscal fourth quarter, a 265 per cent increase from 12 months earlier. The company briefly flirted with a market value of US$2 trillion on Friday, based on global demand for its Graphics Processing Units (or GPUs), the specialised electronic circuitry that AI needs. Nvidia founder and chief executive Jensen Huang trumpeted a “tipping point” for faster computing and AI as the company results were published. Ernst &amp; Young managing partner Selina Short agrees with the sentiment. ▲ It’s just about linking the AI engine to the right data set. “Interest in the technology is being driven from the top down, and boards are instructing their businesses to start piloting, testing and trialling,” Short told The Urban Developer. “It’s our view that the transformation of labour will be universal—in that it will affect nearly all functions—but the degree will vary across sectors and geographies.” In the first instance that means planning and design. Archistar head of generative design Anycie Barakat says the industry is already using AI to read and understand planning rules. “So, if you have a document of planning rules from a local government area, and you want to know what you can do on a particular development site, AI can go through those rules and tell you what you can build,” Barakat said. As a property development platform, Archistar holds thousands of PDF documents on all manner of local and regional planning. Plugged into ChatGPT—the chatbot developed by OpenAI and launched just 15 months ago—a machine can take that data and write a tender document. “I think it depends upon what stage you are looking at,” says Barakat. “I think you can go from the pre-development application phase and concept design stage, which it’s heavily been used in now. ▲ Archistar head of generative design Anycie Barakat: Eventually, AI will go through the rules and tell you what to build. “And there is a lot of room to expand as well, so looking into actually getting DAs signed and construction documentation approved, all the way to rendering, sales and community engagement. “It’s just about linking the AI engine to the right data set.” Barakat, who like Melbourne University’s Pugnale began life as an architect, believes AI can solve one of Australia’s biggest development issues—the time it takes to get from the design stage to design-approval. Three years ago Archistar began working with the Victorian government and the building industry to introduce new self-service digital assessment technology that will make the planning process easier and faster. Known as eComply, the project will use 3D modelling to take various parameters from a municipality’s Local Environmental Plan (LEP), set them against a certain development site and upload the outcome to a state planning department website. “We brought in a number of different builders, surveyors and architects to confirm the technology and make sure that it’s there.” Barakat says. ▲ A digital twin of 300km of Brisbane and surrounds. It has been used to help plan the city’s Cross River Rail project. The model is now in production and being used to assess designs against Victoria’s Small Lot Housing Code. And earlier this month the New South Wales government allocated $5.6 million towards AI to accelerate the assessment of development applications within the state. Archistar is one of a number of companies that’s been asked by the Department of Planning, Housing and Infrastructure to submit “existing, mature, proven and successfully implemented AI technology solutions” to assist councils with assessing and streamlining the pre-lodgement phase. All three eastern seaboard states now have so-called Digital Twin programs, which enable governments to create a digital replica of a region to show existing infrastructure, such as water, sewage, drainage and bridges, on a single platform. Brisbane, for example, has a digital twin of 300sq km of the city, which in its earliest integration was used to help plan the $6.3-billion Cross River Rail project—the biggest public transport infrastructure in the fastest-growing city’s history. It is now being used in planning for the city’s 2032 Olympic Games. ▲ Melbourne University’s Alberto Pugnale is co-director of the school’s Advanced Digital Design and Fabrication Research Hub. Andrew Deveson is a director and co-founder of the Melbourne-based construction company Kapitol Group. In January, he hired the group’s first chief technical officer. “We did that because our challenge as an organisation is to actually get ourselves ready to be able to use AI effectively,” he said. And tomorrow, in a second part to our series, we’ll take a look at how Artificial Intelligence is shaping those with hardhats on their heads and boots on the ground. Our main image for this story was generated using AI, and is the copyright of Nick Moore. The article itself was written by a real human. You are currently experiencing The Urban Developer Plus (TUD+), our premium membership for property professionals. Click here to learn more.

Data Centres
Ralph Nicholson
Tue 27 Feb 24

[+] Juggernaut That Won’t Be Stopped: The Coming of AI

Get the full story.

Arms Race is Over: Era of Pared-Back BtR Rolls Out

Melbourne Value Gap a ‘Counter-Cyclical Opportunity’

Mosman: Sydney NIMBY Suburb Goes to War over Density Reform

Sunland Rises Again, Files Glitter Strip Skyscraper Plans

Construction Industry Rebrand Key to Unblocking Supply

Development Sites: Data-Driven Acquisition Strategies | Melbourne | Tim Hyland | National Strategy Manager - Transactions & Advisory, RPM

In conversation with Matthew Belford | Managing Director, ID_Land

2026 Residential Developer Summit Melbourne | Panel | Anne Jolic, Rotem Rotenberg, David Stone & Julian white

Construction Forecasting Benchmarks | Melbourne | Yew Zhi Tan | Senior Quantity Surveyor, Mitchell Brandtman

Time & Place Plots $181m Surry Hills Adaptive Reuse Project

Arms Race is Over: Era of Pared-Back BtR Rolls Out

Billionaire Racing Titan’s Gold Coast Twin Towers Out of The Gates

Stockland Swaps Warehouses for $2bn Melbourne Data Centre

Adelaide Developer Symposium

Land Lease Summit

Urbanity-26 Conference

The Urban Developer Awards for Industry Excellence

Arms Race is Over: Era of Pared-Back BtR Rolls Out

Melbourne Value Gap a ‘Counter-Cyclical Opportunity’

Mosman: Sydney NIMBY Suburb Goes to War over Density Reform

Sunland Rises Again, Files Glitter Strip Skyscraper Plans

Construction Industry Rebrand Key to Unblocking Supply

Data CentresRalph NicholsonTue 27 Feb 24

[+] Juggernaut That Won’t Be Stopped: The Coming of AI

Get the full story.

Arms Race is Over: Era of Pared-Back BtR Rolls Out

Melbourne Value Gap a ‘Counter-Cyclical Opportunity’

Mosman: Sydney NIMBY Suburb Goes to War over Density Reform

Sunland Rises Again, Files Glitter Strip Skyscraper Plans

Construction Industry Rebrand Key to Unblocking Supply

Development Sites: Data-Driven Acquisition Strategies | Melbourne | Tim Hyland | National Strategy Manager - Transactions & Advisory, RPM

In conversation with Matthew Belford | Managing Director, ID_Land

2026 Residential Developer Summit Melbourne | Panel | Anne Jolic, Rotem Rotenberg, David Stone & Julian white

Construction Forecasting Benchmarks | Melbourne | Yew Zhi Tan | Senior Quantity Surveyor, Mitchell Brandtman

Time & Place Plots $181m Surry Hills Adaptive Reuse Project

Arms Race is Over: Era of Pared-Back BtR Rolls Out

Billionaire Racing Titan’s Gold Coast Twin Towers Out of The Gates

Stockland Swaps Warehouses for $2bn Melbourne Data Centre

Adelaide Developer Symposium

Land Lease Summit

Urbanity-26 Conference

The Urban Developer Awards for Industry Excellence

Arms Race is Over: Era of Pared-Back BtR Rolls Out

Melbourne Value Gap a ‘Counter-Cyclical Opportunity’

Mosman: Sydney NIMBY Suburb Goes to War over Density Reform

Sunland Rises Again, Files Glitter Strip Skyscraper Plans

Construction Industry Rebrand Key to Unblocking Supply

Data Centres
Ralph Nicholson
Tue 27 Feb 24