**3. Discussion**

### *3.1. Can Artificial Intelligence Help Cities Become Smarter?*

Cities are complex organisms and their complexity increases exponentially as they continue to grow [63]. With computational abilities vastly superior to humans, when it comes to ingesting large swaths of data, AI systems are among the core elements of most smart city projects [64].

Other smart technologies such as internet-of-things (IoT) [65], autonomous vehicles (AV) [66,67], big data [68], 5G wireless communication [69], robotics [70], blockchain [71], cloud computing [72], 3D printing [73], virtual reality (VR) [74], augmented reality (AR) [75], digital twins [76], and so on are also transforming our cities [77].

For instance, it is increasingly common to combine machine learning with other emerging technologies to generate advanced urban solutions. Examples include: the use of deep learning and high-performance computing (HPC) for traffic predictions using sensor data [78], incident prediction [79], disaster managemen<sup>t</sup> [80], and rapid transit systems designed to optimize urban mobility systems [81]. Machine learning has also been used with big data technologies and social media for logistics and urban planning [82,83], event detection for urban governance [84], disease detection [85], and identifying the sources of noise pollution at the city scale [86].

Additionally, machine learning has been applied along with distributed computing to improve basic scientific computing operations that are fundamental to urban design modeling methodologies [87]. Moreover, machine learning is paired with IoT for human activity recognition [88], smart farming [89], and developing next-generation distance learning systems [90]. Furthermore, machine learning benefits from data fusion in ubiquitous IoT environments [91], where this creates a potential to significantly enhance AV decision capabilities [92]. Figure 5 lists AI capabilities and their use by domains.

Nevertheless, it is when AI is combined with these technologies that we can really see its big potential to address complex challenges and harness opportunities within our urban environments— given that some ethical issues are adequately addressed. Despite the AI and ethics issue being discussed in academic and governmen<sup>t</sup> circles, so far only limited guiding principles have been produced and legislated [94]. In that regard, the European Parliament's [95] initiative on guidelines for the European Union (EU) on ethics in AI is a commendable but limited attempt, as ethical rules on AI are so far essentially of a self-regulatory nature, and there is growing demand for more governmen<sup>t</sup> oversight.

**Figure 5.** AI capabilities and their use by domains, derived from McKinsey Global Research Institute [93].

### *3.2. What Are the Promises and Pitfalls of Artificial Intelligence for Cities?*

A recent study [96] that evaluated the levels of smartness of Australian local governmen<sup>t</sup> areas advocated for the importance of integrating urban technologies, including AI, into local service delivery and governance, for instance, the use of AI in tasks that enhance environmental sustainability, such as sorting waste for recycling [97]. Additionally, the practice review conducted by McKinsey Global Research Institute [93] discloses projects from across the globe where AI is utilized for achieving UN's sustainable development goals (SDG) (Figure 6). Nevertheless, before AI is implemented on a wider scale, it is important to understand how this technology can contribute to making our cities (and the planet) smarter. Conversely, understanding the pitfalls of AI will enable us to ensure AI delivers the desired outcomes in urban areas and beyond.

**Figure 6.** AI utilization for achieving sustainable development goals, derived from McKinsey Global Research Institute [93].

With the above-mentioned issue in mind, our team in another study [22] evaluated the promises and pitfalls of AI for cities according to the main smart city dimensions of economy, society, environment, and governance [98]. Table 4 below summarizes the key findings of the study.


**Table 4.** Promises and pitfalls of AI for cities, derived from Yigitcanlar et al. [22].

### *3.3. What Are the Ways to Maximize Artificial Intelligence Promises and Minimize Pitfalls?*

The biggest pitfalls of AI-enabled solutions are that they may aggravate the existing socioeconomic disparity [99] and have privacy [100] (for example, increased governmen<sup>t</sup> surveillance during COVID-19) and cybersecurity [101] issues. Most of our cities are already fragile and inattention to how local governments maintain social compacts will only increase their fragility [102]. It is imperative that technological progress does not accelerate the widening of existing fractures, or incubate new sources of fractures, in our cities [103].

Given the fast-paced implementation of AI, it is important that we act now and find ways to minimize the pitfalls of AI while maximizing its promises [104]. Some of the useful actions are presented below.

