Proposing the foundations of scAInce by exploring the future of artificially intelligent, sustainable, and resilient megaprojects

Abstract

Artificial intelligence is software based; megaprojects are large physical infrastructures. We use exploratory research in understanding the merger of both by asking what artificially intelligent, sustainable, and resilient megaprojects are and what their underlying theory is. Thus, the goal of the paper is to define an artificially intelligent megaproject and identify its theory. To do so, we conducted a literature review on the two different topics and exemplified current mergers of AI and megaprojects through two cases. We propose a definition for artificially intelligent megaprojects and suggest consequentially an embedded new theory we call scAInce. We suggest that scAInce is a new research field, that explores the influences and impacts of artificial intelligence on our world. We conceptualize how AI megaprojects as virtual AI futures might enable sustainability in our built environment. It ends with contributing a definition of what artificially intelligent megaprojects are.

HIGHLIGHTS

• Defines AI megaprojects as a synergistic and enduring merger of AI software for large-scale infrastructure planning, design, construction, and management

• Suggests a new research field called scAInce

• Introduces the theory of scAInce as utopian and dystopian visions of futurists that are starting to emerge in the incremental transition from smart to the intelligent city concept

Keywords:

• Mega project
• artificial intelligence
• digital twin
• urban science

Introduction

Megaprojects are increasing in scale, scope, and size with an emphasis on sustainability and resilience. Some researchers, therefore, have argued that AI is useful to manage the complexity of megaprojects. Beyond management, we found that AI is significantly more intrusive to megaprojects culminating in our proposal for a theory we call scAInce. Thus, this paper pursues three interrelated goals: 1) to introduce scAInce as a new research field 2) to define scAInce conceptually and 3) to illustrate artificially intelligent megaprojects as examples of how AI is transforming a long-established science field called megaprojects. Thus, our research questions are:

• What are artificially intelligent, sustainable, and resilient megaprojects?

• What is the underlying theory of artificially intelligent megaprojects and their concepts?

We theorize scAInce and define artificially intelligent megaprojects by taking a positivistic research approach as a logical exploration (Yin 2003) of a system or principle. Exploratory research is constructed to answer ‘what’ questions, according to Yin (2003), and in many cases includes new studies which can be a result of the appearance of new phenomena or insufficient knowledge. We explain the underlying concepts of scAInce by using artificially intelligent megaprojects as examples. In general, a theory consists of principles that are formed to explain things, phenomena, or a group of facts already validated by data. According to Currie and Knights (2003), the positivist approach starts with dilemmas, ideas, or observations. Following Currie and Knights (2003), we analyze the fact and objectives of the study to create a logical way of thinking to explain the subject of the study. We use as examples emerging digital twins. Prediction of prospective behavioral development of digital twins is used as a positivist approach to generate general laws from the standpoint of objective truth (Fisher 2004). scAInce is a novel science, that conceptualizes how and why artificial intelligence influences and impacts our world as we know it. This upcoming revolution, in which every aspect of our lives will be influenced by artificial intelligence, is jointly driven by private companies and governments in the search for ever-greater efficiency, governments in their aspirations towards sustainability, and ultimately, citizens across the world, who adopt, alter, or reject new instantiations of AIs within their everyday worlds (Kassens-Noor et al. 2021). In this paper, we focus on artificially intelligent, sustainable, and resilient megaprojects, because we believe it is in the merger of these complex concepts, in which AI shows potential to enable sustainability. We define artificial intelligence as computational technologies that employ machine learning including deep learning methods that reference human intelligence to address complex decision-making (Berente et al. 2021; Galaz et al. 2021). Furthermore, AI is dynamic and always evolving toward new challenges and computational frontiers. Sustainability is a concept that we intuitively identify as associated with responsibility, the term is bolstered by the ISO 37101:2016 standard, ‘Sustainable development in communities’, created under the precepts of the 2030 Agenda of the Sustainable Development Goals for SDG 11. Sustainable cities and communities promote the responsible use of resources and the environment and the improvement of the well-being of citizens (Florez-Jimenez, Muñoz-Villamizar, and Lleo 2021; Ramirez Lopez and Castro 2020). Resilience is defined as the preparation, response, and adaptation to incremental change and sudden disruptions, so systems survive and prosper (Florez-Jimenez, Muñoz-Villamizar, and Lleo 2021; Ramirez Lopez and Castro 2020).

The paper starts with introducing AI and megaprojects of the past that are conceptualized as AI utopias, continues with the current Epoque in which these utopian ideals confront reality within the incremental evolution of smart city concepts, and third forecasts the transition from smart to artificially intelligent megaprojects within cities and regions by describing two virtual cities currently in development to change mega infrastructures in the real world. To foreshadow our findings, we define artificially intelligent, sustainable, and resilient megaprojects as large-scale, enduring, complex, and dynamically changing software and hardware systems that bridge the virtual and physical worlds. Through the application and testing of new AI systems on virtual megaprojects’ planning, design, construction, and management, sustainable and resilient solutions are developed, tested, and deployed across real-world megaproject lifecycles. Megaproject cycles are subsequently nourished through the continuous application of big data and connected AI systems. Megaproject data and AI-driven computational power enable megaproject-driven real-world applications and sustainable solutions through optimization, redesign, and reconstruction. High-capacity systems facilitate virtual modelling, which identifies those configurations that show the greatest improvements for the sustainability and resilience of real-world urban environments.

A brief history of AI and megaprojects as visionary decades-old concepts

The origins of artificial intelligence can be traced back to the middle of the last century when scientists working on corporate and defense projects created machines designed to mimic the intellectual capabilities of the human brain. Two dominant approaches to AI emerged at the time: symbolic logic model-based learning and neural or machine learning (Haenlein and Kaplan 2019). The intellectual rivalry combined with insufficient computing power and sensor data sidelined mainstream AI initiatives for more than thirty years (Lipson and Kurman 2016). It was not until the early 1990s when few researchers, who had continued as outsiders, were able to make pattern recognition such as reading text and decoding speech possible. But only when breakthroughs of this advanced technology were implemented in banks and postal systems, AI became mainstream. By the mid-2010s the computing power of machines had rapidly increased and enabled the next significant step in AI research: deep learning, in which more than one neural network was woven together to enable machine learning’s predictive capability. Deep learning signified the machine’s ability to be trained (or learn) from millions of examples to correctly identify and classify objects, even if that particular picture had never been loaded into the machine. We are today (2022) at a point, in which machines can handle huge amounts of data and make complex calculations very quickly, or as Arent Hintze (2016) categorizes them, type II robots. AI technologies most frequently used include image analysis (35%), virtual assistants (31%), predictive analytics (29%), machine learning (28%), and natural language processing (26%) (Shuss 2019). AI has and is predicted to develop from being merely reactive machines (no memory and limited futures), to limited memory as transient pieces of memory as type II, to machines developing an understanding that people, creatures, and objects in the world can have thoughts and emotions that affect their behavior, to finally to self-awareness, that can form representations about themselves (Hintze 2016). The AI vision is to become sentient machines with massive computing power and limitless memory, that are conscious of themselves and their surroundings. For example, reconstructing the human brain is an AI megaproject pursued by many.

In parallel to these virtual software development megaprojects, infrastructural megaprojects in the past decades have significantly restructured our built environment in an attempt to increase efficiency in areas such as urban mobility. These megaprojects have traditionally focused on urban challenges with results that have affected cities and their process of urbanization. They were built in response to neoliberalism and globalization within the urban realm (Theodore, Peck, and Brenner 2011) and in doing so, have deep impacts on the relationship between urban developments and their embedded political, social, and economic power relations in the city frequently producing new forms of governance (Moulaert, Rodriguez, and Swyngedouw 2003). Megaprojects have been critiqued as they were significantly plagued by cost overruns, time delays, and not meeting specifications (see Morris and Hough 1987; Flyvbjerg, Bruzelius, and Rothengatter 2003; Samset 2012). But, analyzing 30 megaprojects exceeding $1 billion worldwide, the OMEGA Centre (2012) showed that most megaprojects were actually close to being on time, on budget, and within specifications. The researchers argued that conflicting evidence necessitates a megaproject evaluation framework that judges such projects for how they meet objectives over time and within their context, while frequently facing shifting societal, political, and environmental values. In recent years, new MIPs have become larger, more complex, and more crucial for our societies as they bring not only greater negative impacts but are also more impactful locally and globally as we strive toward sustainability (Fahri et al. 2015; Lehtonen 2019; Dimitriou, Ward, and Wright 2017; Söderlund, Sankaran, and Biesenthal 2017; Samset and Volden 2016).

Sustainability in AI-driven megaprojects within the context of this article embraces sustainability as a core tenet throughout the megaproject’s lifecycle from concept context to sustainable production. A holistic approach to sustainability grounds expectations and measurement throughout the process of project conceptualization, development, delivery, production, and potential winddown to deliver social and environmental sustainability over the extended lifecycle of the project, whether that be a new port, international rail connection, or a new energy generation project (Rakic and Rakic 2015). However, megaprojects have traditionally been rationalized around the need for economic development and improved diplomatic ties with the social and environmental aspects being a second thought. However, China’s Belt and Road Initiative (BRI) incorporates megaprojects, in which many have international partner countries that struggle with significant environmental issues. Xiaolong et al. (2021) investigated the efficacy of incorporating environmental objectives into projects and the use of sustainable project management practices in megaprojects to impact environmental performance in partner countries by looking at BRI projects in Pakistan and China. The authors found that environmental requirements boosted instead of hindered the project performance and sustainable management techniques improved instead of hindered diplomatic (international) relations.

To optimize megaprojects, implementing artificially intelligent systems into large-scale projects has become more frequent over the last decade (Economist Intelligence Unit Limited 2018, 8): AI technologies included image analysis (35%), virtual assistants (31%), predictive analytics (29%), machine learning (28%), and natural language processing (26%). The recent advances in AI fall under the rubric of machine learning, which involves programming computers to learn from example data or experience (Economist Intelligence Unit Limited 2018). Though the implementation of AI poses significant challenges, large engineering and construction firms deploy AIs to mitigate complexity (Greiman 2020). Megaprojects are extraordinarily complex and thus having more sophisticated tools provides increased opportunities for sustainable delivery, so the companies believe. One downside of implementing novel AI systems is that the projects’ risk profiles increase. The more complex the tools, the greater the opportunity for error or malicious behavior. The author highlights the complexity of managing projects with AI and AI initiatives and describes how bias in AI-driven decision-making has the potential for disastrous results. Similarly, citizens can steer AI acceptance on artificially intelligent megaproject operations towards unacceptable outcomes. Anthony Townsend in Ghost Road effectively argues that unbridled consumerism, in places like the U.S. leads to unintended choices about communities – based on our de facto collaboration with the AI industry. The more Door Dash we order the more we will need AI-enabled mobility, cityscapes, and housing (and not for environmental reasons but to keep our Amazon orders coming). Our cities, thus, will change to accommodate our fulfillment and may result in ghost kitchens.

Utopian and dystopian visions for AI and megaprojects collide consistently. Like AI advocates, megaproject protagonists similarly embrace a narrative of when cities implement either one, or both combined. AI provides cities with a competitive edge in the world city system and connects their AI implementation to growth, development, a better quality of life, greater wealth, and continuous prosperity. While single projects, like the smart mall, smart museums, or Masdar City, have frequently caught our attention as successful marriages between new AI solutions and megaprojects, they are built upon an utopian idea. These projects take an empty space to construct an AI future that allows entry to those who pursue that dream largely disregarding the existing built and natural environment, or the society that is supposed to be living in it. If we disregard important push and pull forces of AI in a society like migration pressures, we might miss important trends or discard alternative futures to the ones we wish for (Kassens-Noor and Hintze 2020). As such these utopian physical ideals have generally failed in that they were not accepted by society at large or could not even be implemented as cities are not blank slates. The smart city concept, in contrast, approaches the incremental introduction of technology into cities in an attempt to tackle ever-increasing sustainability and climate challenges. Indeed, smart cities have emerged as solutions to urban sustainability using technology innovations to transform the global society.

AI visions meet reality: the evolution of the smart city concept

Smart-city megaprojects have been envisioned as livable, healthy, and innovative, assuming urban sustainability is achieved through technological advances. In lockstep with new AI developments smart city technologies are implemented in the hopes that incremental introduction over the long run will gain public acceptance (Kassens-Noor et al. 2021). Too aspirational projects, or alternatively phrased in-reality grounded artificially intelligent megaprojects, however, have been tainted with failure like the Sidewalk labs in Toronto. A host of recent sustainable urban megaprojects in the Middle East have been questioned on social and environmental grounds (Rizzo 2017). The future of AI though keeps being envisioned brighter: AI-driven solutions are seen as bold and as having the computational power to address climate change, the single greatest existential threat to many cities (Yigitcanlar, Desouza, et al. 2020). Yigitcanlar, Desouza, et al. (2020) evaluated 93 smart city articles that feature AI, of which one-third are environmentally focused. At this time, AI solutions are deployed as point solutions or as a part of a strategy that may tie into sustainability. However, carbon neutrality today does not necessarily produce sustainable outcomes for future generations.

The incremental approach to smart city implementations bears significant threats to data security and privacy in smart cities (Ismagilova et al. 2022). The sources of these threats are embedded in the setup of our mobile devices and services, the smart cities’ infrastructure and technical architecture, the power systems, smart healthcare solutions, the security and privacy frameworks that govern the cities; as well as algorithms and protocols pose operational threats for smart cities and social media fostered in and about smart cities (Awad et al. 2019). Given the penetration of these different smart device technologies, privacy and security can be easily compromised due to frequent and connected interactions between people, their devices, and local sensors (Elmaghraby and Losavio 2014; Picon 2019). Given the incremental implementation approach by a variety of actors, the different components of smart cities as mentioned above face a countless number of security threats (Baig et al. 2017; Kitchin and Dodge 2019).

Similarly, artificial intelligence algorithms are incrementally introduced to smart city projects. –Subsequently, Yigitcanlar, Butler, et al. (2020) posit that artificially intelligent cities are replacing smart cities and have the capacity to safeguard residents from disasters, pandemics, and catastrophes. They further argue that AI is becoming the new ‘electricity’ which will power all intelligent (smart, redundant, etc.) systems. AI enables better and faster decisions which leads to safer residents and provides avenues for negative carbon growth to ensure the viability of future generations.

However, visions of technology have promised a future, which they have been failing to deliver for decades (Kassens-Noor and Darcy 2022). Thus, the future of artificially intelligent cities might look significantly different than today’s paradigms. Let’s take transport as an example. For the history of humankind, transport served as a means to an end. We wanted to reach a destination, and transportation engineers and planners have traditionally either minimized distance or time. The distance parameter was and remains straightforward, it would not change significantly given new infrastructure is costly and existing environments are built to last for decades. In contrast, new inventions to minimize travel time for individual drivers have changed with the entry of IoT over the past decade. Apps like Waze, Uber, and Lyft uprooted the transportation industry calculating minimal travel time by connecting service providers with the individual or by connecting other app users, in which the embedded AI algorithm predicts ever-changing travel time and charging end users based on current demand that might vary by time of day, weather, and availability of drivers. With the advent of artificial intelligence as megaprojects in smart cities including autonomous cars, the private industry’s entry into the time & distance transport paradigm is slated to add a third dimension that never existed in time/distance optimization: maximizing profit. While to date maximizing profit has been based on available demand, once AVs drive on our roads, routes can also be plotted to maximize spending by the customer or rider of the AV – this may be prolonging in-car time to let riders finish a movie, or a different route to pass by shops or restaurants that they frequent. This new paradigm signals a shift in the transport sustainability paradigm as well, emphasizing once again the economic dimension over the cost of the social and environmental one.

From smart to artificially intelligent megaprojects and the foundations of scAInce

Given the slow evolution towards artificially intelligent cities as described above, we hereby conceptualize the theory of scAInce as utopian and dystopian visions of futurists that are starting to emerge in the incremental transition from smart to intelligent cities. Megaprojects and other infrastructures are the early adopters and enablers of big data projects that are legible to machines. An artificially intelligent megaproject enables big uncompromisable goals of sustainability and resilience but allows the artificial intelligence system flexibility in how to achieve them. The AI megaproject system continuously collects data and analyzes evolving patterns to apply both in an endless loop towards more sustainable solutions. Thus, an AI system is dynamic and ever-changing and can with a single change in an algorithm enable transformational sustainable change. Referring to our transport example above, a different AI algorithm can optimize travel time over distance to minimize emissions given the dynamic contexts of city traffic. In doing so, the AI fundamentally changes mobility patterns whereby the only fixture remains the city’s built environment.

But the AI megaproject’s ability for change also has its challenges because AI remains a black box. As an autonomous system, which Norton (2021, 226) defines as ‘freedom from external control or influence’, humans will have to agree and be comfortable with holding limited – if any - control over the development and usage of artificially-intelligent megaprojects. The use of AI in intelligence analytics has posed many challenges including data analytics, the integration of human and artificial intelligence, and moral and ethical concerns. For example, AI has an inherent bias: AI is very good at interpreting data from the past but is still limited in its ability to envision the future. Lessons from infrastructural megaprojects can be used to overcome this mistrust as megaproject risk has to be strategically managed. In recent literature (Ward 2020), strategic planning risk, defined as the treatment of risk as a response to uncertainty over the longer term, is identified and quantified as an increasing threat to megaprojects as we enter the expected impacts of climate change and pandemics. The key is to better identify and manage that strategic risk, one of which is the challenge for humans to accept a life form that is going to be smarter than themselves.

To enable artificially intelligent megaprojects, mistrust in technology must be overcome. However, skepticism is already widespread in the phase II AI transition phase (Hintze 2016) and not on a visionary concept when AI supersedes human capacity and can think for itself (Type IV). For example, both Norton (2021) and Townsend (2020) critique autonomous cars, which are likely to be the first widespread use of AI robots in our lifetime as the ‘mindless extension of the humans who wrote the program that governs its decision’ because ‘a fully automated car is not actually smart, intelligent, or autonomous’ (Norton 2021, 226). Similarly, Townsend (2020, 227) advocates for the ‘need to expose the code that controls these menacing machines before they hit the streets’. Both describe the code as the true battleground of the future, where our assumptions about efficiency, safety, and fairness will be inscribed into algorithms. Both authors fear that malleable code makes the system vulnerable and error can be introduced with a single computer stroke. Megaprojects as urban planning projects similarly attract similar critique that is decades-old in the planning literature, in which utopian ideals of perfect form ultimately enable perfect functions (Howard 1965). Autonomists today believe that the mass adoption of autonomous vehicles will unleash a virtuous cycle of social and economic progress (Townsend 2020). But many planners have argued that precisely that assumption is unrealistic, as people are the centers of the city and thus change comes from within (Jacobs 1961). A likely hurdle to the aspirational introduction of AI by governments and private companies is founded in the research on megaprojects: the sheer introduction of megaprojects into the public eye can create civic mistrust due to their magnitude and consequent effects on large areas of a city, their enormous economic costs, and their massive environmental impacts (Brenner and Theodore 2005; Pitsis et al. 2018).

The risks of AI and megaprojects

Taken together, the risks associated with both, artificial intelligence and megaprojects, can be reduced in artificially intelligent megaprojects. Stakeholders can herewith address the overall risk in its planning stage, which is intrinsically linked to megaproject complexity, to reduce the risk for the project and its performance (Floricel, Michela, and Piperca 2016). As introduced above, the main problem with the smart city concept is that code is introduced incrementally via individual unconnected devices and technologies (Ismagilova et al. 2022; Awad et al. 2019; Baig et al. 2017; Kitchin and Dodge 2019). Because the risks are addressed individually and not upfront as an interconnected system, the smart city bears many risks. For example, automated cars are tested as experimental approaches of trial and error. That introduction has created safety hazards, and caused deaths. Another example is when AI created chatbots that turned into genocidal maniacs in a matter of hours (Ohlheiser 2016). This is exactly the project context, which Ward et al. (2019) referred to when emphasizing that these contexts are major sources of project risk. Recent developments, like the pandemic, created new contexts bearing additional risks. AI and new technologies were pushed online that were not proven to be safe, i.e. when Zoom reported a significant breach and loss of privacy (FTC 2020). With new MTPs on the forefront to be deployed as economic stimuli to uncertain economic events (Anderson 2021) that are envisioned to link AI investments and growth (Bugin et al. 2017), managing the megaprojects’ risk becomes ever more important.

Already a decade ago, Brenner and Theodore (2005) identified planning as the key to managing the strategic risk for the megaproject: allow megaprojects to prosper through widespread social and political support. Yet, the trend in megaproject delivery went the opposite path. Consequently, civil society perceived megaproject conception as filled with secrecy which has contributed to mounting opposition to the development and implementation of UMPs. Harris (2017) observed ‘introverted modes of governance that circumvent local planning frameworks, traditional democratic channels of participation, and accountability’. To break this impasse, Dimitriou et al. (2014) advised striking a balance of adopting openness when aligning strategic objectives and transitioning to project closure when the project needs to be delivered.

Artificially intelligent sustainable and resilient megaprojects must be planned and delivered in the public eye. AI infrastructure and algorithms can optimize sustainability and resilience striking the delicate balance in managing megaprojects between private companies’ interests to deliver cost savings and the government to guard the public interest. Walz and Firth-Butterfield (2018, 2019) in particular highlighted that AI systems pose major risks including the causation of damages, loss of privacy and personal autonomy, information biases, and susceptibility to manipulation of AI and autonomous systems. Directly, AI may lead to 21%-38% job loss (Berriman and Hawksworth 2017, 1; Lin, Abney, and Bekey 2011, 942), loss of social interactions (Groth, Nitzberg, and Esposito 2022), digital dementia (Spitzer 2014), and exploitation of the human race (Polonski 2017; Smith 2018; OECD. 2017). Thus, artificially intelligent, sustainable, and resilient megaprojects should adopt an open science approach, in which citizens and residents enable and are enabled to participate in the planning, design, construction, and management of AI megaprojects to achieve sustainability and resilience.

The definition of an AI future

We define artificially intelligent megaprojects as large-scale enduring software and hardware systems bridging the virtual and the real world. What makes them sustainable and successful is civic buy-in and control. In the following section, we are introducing two artificially intelligent megaprojects in the making, that can overcome the strategic risk that has plagued many AIs and mega projects, and combined can become an artificially intelligent, sustainable, and resilient megaproject.

AI-driven digital megaprojects promise to deliver new and improved physical experiences for urbanites. In the following section, we introduce two AI projects that are being developed as digital twins. The first re-envisions living in a city that has yet to exist allowing new ideas and forms of living to evolve that are in the physical world constrained by the built environment. The second models an existing city in the hopes that the digital twin enables an accurate replication of real-world complexities by modelling individual behaviours of avatars through interactions among different transport systems and technologies.

A virtual city for future living

The idea of Spectra is to rethink infrastructure to generate massive efficiencies. So far, the efficiencies required the deep reorganization of urban space in the virtual reality city. This reorganization of urban space so the hypotheses of its developers will create or make room for new types of urban life that are more sustainable. The Spectra business owners and coding architects propose a completely new technologically advanced city, which residents across the world will collaboratively design using a digital twin experimentation space. Rzepecki et al. 2022 envision a world, in which residents engage in a virtual environment for their work, community, and personal lives by collaborating on projects, gaming, or managing their assets. The concept is unique in that the digital community will inform a yet-to-be-build physical city designed for and by its community members through Web 3.0 applications. Stakeholders will contribute to the city’s design, implementation, and investment. As a planned city, founders of the future city anticipate the process will lead to practical solutions that reimagine how residents build and navigate urban spaces. The community will likely be financed with native tokens, which can be earned through a variety of activities including mining, gaming, building the network, staking, and engaging with the community. Leveraging a virtual world with AI-driven features to propose remedies for issues that ail the physical world such as unsustainability, environmental concerns, social disenfranchisement, health crises, and the dominance of global corporations is new. The digital/physical structure the developers are proposing is a novel way in which to incorporate citizen participation in the virtual conceptualization and testing of new physical urban environments and social structures (Figure 1).

Figure 1. Urban life in the virtual city Spectra © Spectra and Numena. Reproduced with Permission.

A vision of an artificially-intelligent megaproject: Darmstadt’s digital twin

The Technical University of Darmstadt (Germany) and Deutsche Bahn AG have embarked upon an AI-driven virtualization of the German Rhein-Main Region in greater Frankfurt to improve regional travel efficiencies. At the heart of the solution is a vision for an AI-driven simulation platform of the region to model regional rail transport demand and intermodality options (Figure 2).

Figure 2. Digital twin of Darmstadt.

DB AG and TUD jointly developed SMD (Simulation platform for Mobility, Transport, and Traffic Darmstadt), a new modular simulation platform with four modules that facilitate real-time planning for system administrators and travelers. Module 1, the SMD Virtual World, represents virtual stations and rail lines based on the physical rail infrastructure and operation model known as Eisenbahnbetriebsfeld Darmstadt (EBD). Real data of settlements in the Frankfurt RhineMain region were used as reference points so that the virtual world simulates synthetic populations on transport networks for bicycles and pedestrians, rail, and road traffic. Module 2, the SMD demand modelling, incorporates an activity-based approach to model realistic individual travel behaviors on a highly disaggregated spatial resolution at the household level. This demand simulation stores the trips of all travelers in the simulated world in a travel database. Module 3, the Passenger Trip Execution module, also called the Master module, integrates the SMD demand model with traffic flow simulations. Its main functionality is to run each scheduled trip and estimate its current status by communication among the other modules. During real-time simulations, the automatically system implements the replanning process in case of delays. Thus, the trip sequence is no longer determined exclusively by the demand module. Instead, travelers can update their traveler information each time before starting a new trip and weigh the decision of whether to continue with the original or an alternative travel option. Module 4, the traffic flow simulations visualize individual trips and their interactions in an agent-based approach. The modules are divided into different transport systems and sub-modules, representing their infrastructure and operations. In particular, the railway system has its module separated from the public transport system, controlled and monitored by an integrated Traffic Management System. Together, the four modules provide optimization opportunities for one of the business mobility markets in Europe (Figure 3).

Figure 3. Modular structure of digital twin.

Conclusions

Every megaproject has its own goals, mission, economic, social, financial technical, and legal environments to consider and the use of AI should be carefully examined on a case-by-case basis. The conceptual architecture of sustainable and resilient AI megaprojects is synergistic and an enduring merger of AI software for large-scale infrastructure planning, design, construction, and management rooted in the principles of sustainability and resilience. As the two AI megaprojects in the making examples show, artificial intelligence is a conceptually compelling addition to the development and management of megaprojects that seek to be resilient and sustainable systems and communities.

Yet, both approach public involvement very differently. While Spectra cities envision a co-design and development, the digital twin remains hidden from public view except demonstration videos of what the simulation could develop. Given the success criteria of previous megaprojects, we urge more transparency in the use of AI with all involved project stakeholders. As AI megaprojects are starting to evolve, we, as a society, blindly apply the concepts of computational intelligence in a rush of deterministic efficiency and utopianism. We need to sustainably explore and apply the technologies in association and consultation with the intended beneficiaries. Smart city developments have shown repeatedly that once AI works in the real world, clashes are to be expected. To effectively overcome the expected hurdles in development, research on the linkage between developing technologies and their contribution to sustainability is essential. To identify and measure these linkages, researchers could analyze past megaprojects (see for example Xiaolong et al. 2021), could implement pre- and post-studies in yet-to-be-developed megaprojects like though with a more encompassing scope and long-term view, and how this would be measured in a real project situation. Throughout the project’s lifecycle sustainability measures should be continuously tracked. In the planning stage, the sustainability measures have to become an integral part of the megaproject’s development. Modelling that megaproject in the real world enables a prediction and optimization of its sustainability gains.

Disclosure statement

No potential conflict of interest was reported by the authors.