Mobility and transportation sector is a major contributor to the climate crisis. In 2022, road vehicles were the largest contributor to transportation-related emissions, responsible for 74% of global CO2 emissions in the sector (Carlin and Arshad 2024). Between 2019 and 2050, global demand for passenger transport is projected to nearly triple, rising from approximately 44 trillion to 122 trillion passenger-kilometers (ITF 2019). Consequently, pollution and climate crisis attributed to the automotive sector can be expected to increase as the automotive sector lies heavily reliant on high-carbon fuels with 95% of the energy sourced from oil (European Environment Agency 2022).

Each year, 1.3 million people lose their lives in road traffic crashes, making road traffic injuries a global health crisis. Low and middle-income countries (LMICs) bear the brunt, accounting for 90% of these fatalities (European Commission 2024). Research often highlights that human factors contribute to approximately 95% of crashes, while infrastructure and vehicle factors play a role in around 30% and 10% of cases, respectively. “Human factors” encompass various elements, including risky driving behavior, speeding, driving under the influence of alcohol or other psychoactive substances, distraction, and failure to use protective equipment. Additional key factors include driver fatigue, traffic rule violations (such as violating traffic light, illegal crossing, or overtaking), infrastructure deficiencies, and inappropriate speed limits (ibid).

Ensuring sustainability and safety in the automotive industry is reliant on the use of technology and new fuels that carry a lower climate footprint. Artificial Intelligence (AI) is expected to become a critical enabler of sustainability goals in the transportation sector, including elimination of 43,000 fatalities in the United States of America alone (ITSA 2023). By offering sustainability, efficiency and equity, AI can be an enabler of the future of mobility. Advancements in AI technologies, including artificial neural networks and genetic algorithms, have increasingly been applied within the domain of urban transportation and mobility (Lungu 2024). The integration of AI-driven intelligence into transportation systems is driven by the need to adapt to dynamic conditions while enhancing the efficiency and sustainability of transportation networks. Genetic algorithms play a critical role in optimizing vehicle routing by considering variables such as travel time and associated costs (ibid). Enhancing the efficiency of traffic management systems has implications beyond road networks, extending to rail transit and contributing to broader national efforts to reduce emissions and promote sustainable transportation (ITSA 2023). A key advantage of AI lies in its ability to process and derive meaningful insights from unstructured data. While transportation data is becoming increasingly accessible from a diverse array of sources, many transportation agencies face challenges in extracting actionable intelligence from raw datasets, and AI can serve as a crucial intermediary, transforming vast amounts of unstructured data into coherent insights by systematically categorizing, analyzing, and identifying trends at a granular level (ibid). The availability of high-quality data created by sensors, coupled with substantial computational resources, enables AI systems to learn, adapt, and operate with enhanced efficiency and accuracy. This enables AI to infuse cognitive intelligence in the mobility space and enhance the quality of automotive services for the riders.

However, AI adoption remains highly uneven, with LMICs lagging significantly despite their need for AI-driven development. Existing research predominantly focuses on advanced economies, neglecting LICs and exacerbating global technological disparities (Khan et al. 2024). This oversight contradicts principles of distributive justice and global equity, necessitating an exploration of AI’s role in LICs, a theoretical framework for AI catch-up, key areas of AI application, and strategies to bridge the AI gap (ibid). LMICs with strong foundational capabilities may benefit from leapfrogging strategies, while those without such foundations may find learning and acquisition approaches from absorptive capacity literature more applicable. With superior digital infrastructure, the high-income countries and wealthier developing countries are better placed to capture economic value from AI (Schellekens and Skilling 2024), as evidenced by International Monetary Fund’s AI Preparedness Index (IMF 2024). This could translate to dominance of wealthier nations in the high-value sectors as they use AI to enhance their innovation, while the poorer nations struggle to match pace (Schellekens and Skilling 2024). The concentration of both innovation and investment in AI technologies primarily within advanced economies has led to a situation where most emerging economies, excluding China, primarily serve as importers of digital production technology systems developed in the Global North (Anzolin et al. 2024). This dynamic exacerbates existing technological disparities and further widens the gap in capabilities between nations. As a result, developing countries are largely confined to the role of technology consumers, limiting opportunities for learning, knowledge accumulation, and innovation in advanced digital production technologies (ibid). In such circumstances, AI’s role in democratizing smart, safe and clean mobility evenly across the world warrants investigation.

Existing literature offers little insights into the applicability of AI-powered mobility in Global South region, especially in the low-and-middle-income countries. Despite literature existing on the challenges of using AI in technologically disadvantaged nations, the author could not locate any literature that consolidates the factors to evaluate readiness of nations to deploy AI in mobility sector. Deployment of AI-powered mobility requires a robust digital infrastructure and relies on several factors determining the feasibility of AI for cleaner and safer mobility. This paper aims to explore the readiness of the Global South nations for AI implementing approaches in mobility and transportation, and the potential of AI in democratizing access to smart mobility worldwide.

1. Methodology

This paper borrows from the theoretical framework proposed by Costa et al. (2024) that is composed of four fundamental components to be accounted for to assess the effectiveness of AI in any scenario: accessibility, which is essential for the democratic implementation of AI; affordability, ensuring widespread access to AI technologies; usability, a key factor in AI adoption; and ethical regulations, which promote the responsible development and use of AI (Costa et al. 2024). The paper investigates the presence of these components in five different use cases of AI in smart mobility, including:

1. Route optimization using AI for optimizing fuel consumption
2. Advanced Driving Assistance System (ADAS) for enhancing safety in transportation
3. Shared mobility solutions for reducing per capita emissions
4. Smart charging infrastructure for electric vehicles to promote cleaner fuels.
5. Autonomous vehicles for connected and safer mobility

By understanding the gaps existing in the LMICs, the paper argues the following research question:

• Can AI democratize the access to cleaner, smarter and safer mobility across the world equitably?

2. Analysis

Five use cases of sustainable and safer mobility are assessed in this section, which contribute to reduction of GHG emissions and fuel wastage, in addition to promoting safe mobility using AI and connectivity in mobility landscape.

2.1. Route optimization and smart traffic management

Effective route optimization for road transport can reduce fuel consumption and greenhouse gas (GHG) emissions by identifying the route covering all the points through which a vehicle is supposed to pass. Using complex algorithms, AI platforms identify the most cost-efficient and fuel-efficient route by factoring in traffic patterns, fuel consumption and road permissions (Bulusu 2023). This enables fleet management companies to adhere to the strict emissions regulations, such as the one introduced in the United States. With its advanced capabilities in data analysis, decision-making, and automation, AI plays a critical role in optimizing logistics operations, including transportation routing, production scheduling, and delivery management (Mohsen 2023). By leveraging real-time data processing, AI algorithms can generate insights and predictive analytics that enhance the overall efficiency of urban logistics systems (ibid). AI-driven traffic signal control systems can dynamically adjust signal timings in response to real-time traffic conditions, thereby mitigating congestion and facilitating the smoother movement of delivery vehicles (Zhang et al. 2021). Additionally, AI-based adaptive traffic management strategies can continuously analyze and respond to evolving traffic patterns and delivery demands, enabling the optimal routing of vehicles to enhance efficiency and reduce operational costs.

Route planning and intelligent management of traffic relies significantly on availability of real-time data on traffic conditions. For example, the Internet of Things (IoT) enhances AI by supplying real-time data from sensors embedded in urban infrastructure, including traffic conditions, vehicle locations, and environmental factors (Mohsen 2023). This data enables AI algorithms to make informed decisions, fostering a seamless integration of digital and physical systems. Over time, advanced data analysis and pattern recognition improve predictive capabilities, allowing for precise interventions in case of operational inefficiencies (ibid). Smart cities implementing AI for route optimization need a robust digital infrastructure comprising of deep learning networks for automated decision-making for traffic signal timing, emergency response routing and power grid load balancing (BP3 Global 2025). Real-time data processing relies on edge computing infrastructure and cloud-based analytics platform, in addition to 5G networks for high-speed low-latency data communication between autonomous vehicles and smart traffic systems.

Several cities in the technologically advanced countries have implemented AI-powered smart mobility systems. For example, Singapore leverages its high internet penetration and extensive IoT sensor network to monitor traffic, pedestrian movement, and environmental conditions (Santilli 2023). This data optimizes traffic flow, enhances public transportation efficiency, and reduces congestion through AI-driven smart mobility solutions. Barcelona employs IoT and AI in smart parking systems, using real-time data to guide drivers to available spaces, improving parking efficiency, reducing emissions, and enhancing urban mobility (Mohsen 2024). Tokyo integrates autonomous vehicles into its dense urban landscape, while London utilizes AI-powered congestion pricing for traffic management (ibid). Shanghai has adopted AI-driven smart buses, autonomous delivery vehicles, and real-time traffic integration to improve urban logistics. Karimipour et al (2021) found that optimized routes led to GHG emissions reduction by 62% for trucks in a study in Blacktown City of Australia, showing intelligent navigation systems as enablers of sustainable mobility. For example, UPS – the global shipping & logistics solutions company – has deployed On-Road Integrated Optimization and Navigation (ORION) software that offer optimized routes to fleet operators for reducing emissions, carbon footprint and fuel consumption (Bulusu 2023). The AI platform has enabled UPS to optimize routes to avoid traffic congestion and low-emission zones, while enabling the company to reduce fuel consumption exceeding 10 million gallons, thereby eliminating over 100 million miles of unnecessary driving. Consequently, this has led to an annual decrease of approximately 100,000 metric tons of CO₂ emissions (ibid). Use of route optimization, therefore, helps to restrict pollution by reducing the travel distance of GHG emitting vehicles and relies on a robust compute and telecommunication infrastructure for real-time traffic analysis.

2.2. Advanced Driver Assistance Systems (ADAS)

Numerous safety-focused advanced driver assistance systems (ADAS) have emerged in the automotive market to enable vehicles with smart cruise control and object avoidance. ADAS helps with enhanced perception of the driving environment and decreased human errors while driving (Masello et al. 2022). AI technologies help with collision warning, driving control assistance, collision intervention, and parking assistance by reducing human error and sensing the environment to anticipate hazards with higher accuracy compared to the human eye (Fagnant and Kockelman 2015). Application of AI for environment analysis and predictive driving assistance can enhance safety for drivers during the situations most prone to traffic accidents. For example, Masello et al. (2022) reported that high concentration of road accidents in the United Kingdom (UK) occurs during a small set of contextual conditions, such as on rural roads and motorways in the dark conditions. The research showed that ADAS would reduce road accident frequency in the UK alone by 23.8%. Considering the climate poly-crisis, the role of ADAS becomes prominent to enable predictive optimal energy management and improve vehicle fuel economy (Tunnell et al. 2018). The significance of fuel economy in climate action lies in its inverse proportionality to energy footprint, and reduction in energy consumption leads to reduced GHG emissions and air pollution (ibid). Moreover, reduced traffic accidents translate to lower GHG emissions as traffic congestion resulting from roadway accidents is a significant contributor to carbon dioxide (CO₂) emissions. In general, more efficient and timely responses to traffic accidents can mitigate congestion and, consequently, reduce CO₂ emissions. Chung et al. (2013) determined that the average CO₂ emissions associated with a single freeway accident amount to approximately 398.34 kg. Additionally, Wang et al. (2024) established a statistical correlation between the excess CO₂ emissions generated by highway accidents and the effectiveness of various accident management strategies.

ADAS designed for active safety rely on a diverse array of sensors, including Light Detection and Ranging (LiDAR), Radio Detection and Ranging (RADAR), ultrasonic sensors, and various imaging cameras (Tunnell et al. 2018). Advanced ADAS solutions can use over 12 cameras for 360-degree visibility and adherence to the driving lane (Nagpal and Cohen 2022). AI plays a crucial role in computer vision, facilitating the identification of the surrounding environment, with object classification in video streams representing a significant research focus within ADAS (Tunnell et al. 2018). Among the available sensing modalities, camera-based approaches are particularly effective in discerning the type of object that a vehicle is approaching (ibid). The general framework for object detection in ADAS encompasses multiple stages, including image acquisition, pre-processing, segmentation, object detection and tracking, depth estimation, and system control. Pacheco-Capitaine (2023) said that a primary objective of ADAS is to enhance hazard recognition and, accordingly, to either issue an early warning or directly intervene in the vehicle’s driving behavior. To achieve these functionalities, ADAS must accurately classify road types and detect potential disturbances with a high degree of precision (ibid). The integration of AI within ADAS enables the processing of multimodal data from both sensors and cameras, facilitating decision-making in dynamic driving scenarios. This AI-driven approach ensures that appropriate actions are taken in response to environmental inputs, thereby enhancing vehicular safety and performance (ibid).

The growth of ADAS-enabled vehicles has been witnessed in several countries, as in 2021, one-third of new vehicles sold in the US, Europe, China and Japan had ADAS features (Nagpal and Cohen 2022). The New Vehicle General Safety Regulation (GSR2), formally designated as Regulation (EU) 2019/2144, has introduced updated minimum performance standards for motor vehicles within the European Union, mandating the integration of several ADAS (Wood 2024). In the mass-market automotive sector, original equipment manufacturers, such as General Motors, are anticipated to expand hands-off driving capabilities to urban roadways in North America, offering this functionality as an optional feature in premium vehicle models (Narasimhan 2024). In China, select OEMs have already introduced city-assist technologies, such as Navigation on Autopilot, a hands-on driver assistance system, with the potential for future enhancements to hands-off operation via over-the-air software updates (ibid). Furthermore, Level 3 automated driving (L3 AD) has been legalized for highway use in Germany, China, Japan, and specific U.S. states (ibid). This regulatory trend is expected to extend to additional regions, including Canada, the United Kingdom, France, and Sweden, where the legalization of automatic lane-keeping systems and L3 AD on highways is anticipated in the future (ibid).

2.3. Shared mobility

Shared mobility enables users to access vehicles or transportation services on a short-term basis, either for a fee or at no cost. In addition to traditional models such as carpooling, the past two decades have witnessed the emergence of various innovative shared mobility solutions, facilitated by advancements in digital technology. These include business-to-consumer (B2C) car sharing, peer-to-peer (P2P) car sharing, bikesharing, and scooter/moped sharing (Shaheen et al. 2016). The environmental benefits of electric shared mobility are twofold. First, electric propulsion systems produce significantly lower carbon dioxide emissions per kilometer compared to conventional internal combustion engines (van Brecht 2022). Second, shared mobility services contribute to a reduction in overall vehicle numbers (ibid). Empirical studies indicate that a single shared vehicle can replace up to 23 privately owned cars, and shared mobility users tend to accumulate fewer kilometers per vehicle compared to private car owners (ibid). A study conducted by the Organization for Economic Co-operation and Development (OECD) assesses the potential impact of widespread shared mobility adoption on the carbon footprint of urban transportation (Tikoudis et al. 2021). By modeling the modal distribution and aggregate passenger transport emissions across 247 cities in 29 OECD member states from 2015 to 2050, the study provides insights into the environmental implications of these mobility trends (ibid). The findings suggest that, once widely implemented, shared mobility services could yield substantial ecological benefits, with the potential to reduce passenger transport emissions by an average of 6.3% (ibid).

Advanced machine learning models leverage real-time traffic data, historical trends, and user behavioral patterns to generate precise hourly demand forecasts. Neural networks are particularly effective in identifying both temporal and spatial dependencies in ride demand (Numalis 2025). By examining interregional correlations, these AI-driven systems achieve highly accurate predictions of urban transportation needs. This predictive capability enables companies to strategically allocate idle vehicles, thereby optimizing fleet utilization, minimizing passenger wait times, and efficiently aligning supply with demand (ibid). For example, Uber’s Marketplace employs artificial intelligence to enhance dispatching, demand forecasting, dynamic pricing, and user personalization. AI-driven models predict fluctuations in supply and demand by incorporating external variables such as holidays, major global events, and meteorological conditions (Trotolo et al. 2024). Advanced deep learning techniques, including long short-term memory (LSTM) networks, facilitate the anticipation of future market dynamics and support real-time decision-making. According to Uber’s official blog, as of November 2017, the company’s AI-powered dispatch system executed approximately 30 million match-pair predictions per minute, factoring in distance, estimated travel time, and traffic conditions to optimize trip allocation (ibid).

AI is also integrated within shared mobility services for damage assessment and management of shared vehicles, thereby enhancing operational efficiency and responsiveness. AI-driven solutions significantly reduce the time and resources expended on traditionally labor-intensive and repetitive manual inspections while simultaneously safeguarding an operator’s most valuable assets (). By leveraging AI to analyze photographs submitted by customers, the system compares the vehicle’s exterior condition against an extensive database of pre-existing images—both damaged and undamaged—to accurately detect and differentiate between new and prior damage. These images can be systematically archived and retrieved based on damage classification, specific vehicle areas, or license plate identifiers, thereby streamlining damage documentation and tracking. For Leo&Go, this advanced AI-powered system has demonstrated remarkable efficiency, automating the evaluation of approximately 97% of all customer-submitted images and thereby substantially reducing the need for human intervention in the damage assessment process.

2.4. Demand prediction for electric vehicle charging

Electric vehicles (EV) replacing internal combustion engine (ICE) vehicles can eliminate tailpipe emissions and help to tackle the climate poly-crisis. Having a fully electric future of mobility, with EV efficiency matching today’s best models, could save electricity equivalent to the annual consumption of 21 million homes (Huether 2024). EVs convert over 73% of battery energy into motion, whereas ICEVs in urban settings may achieve only 12% due to heat losses (ibid). The transition to EVs will require many drivers to rely on the electrical grid, necessitating upgrades to transmission and distribution systems. Predictive intelligence will be crucial in managing grid demand and ensuring infrastructure stability (ibid). AI serves as a promising technology to scale the development and deployment of EV charging systems.

AI algorithms optimizing schedules for EV charging based on energy demand, grid capacity, and user preferences. Machine learning techniques facilitate predictive maintenance of charging infrastructure, enhancing reliability and efficiency (Balakumar et al. 2024). Smart charging systems integrate AI and the IoT to dynamically manage charging sessions, optimizing energy consumption while minimizing grid impact. AI-driven analytics enable real-time data processing, empowering grid operators and EV owners to make informed decisions regarding charging behavior and energy usage (ibid). By analyzing real-time power demand, grid capacity, and user behavior, AI enhances charging strategies, ensuring EVs charge during optimal periods—when renewable energy is abundant, and electricity costs are low—thereby reducing both expenses and environmental impact (Theissler et al. 2021). Additionally, machine learning-based predictive maintenance detects potential infrastructure issues before they arise, minimizing downtime and improving overall system reliability. AI has the potential to transform the deployment of EV charging infrastructure by optimizing site selection and energy management. By analyzing data from traffic patterns, population density, existing charging networks, and projected EV adoption rates, AI can identify the most strategic locations for new charging stations (EV Industry Summit 2023). This data-driven approach enables transportation planners to minimize installation and maintenance costs while maximizing accessibility and efficiency (ibid). AI enhances energy utilization by continuously monitoring electricity demand, ensuring efficient operation of EV charging stations. By preventing energy waste and mitigating grid overload during peak demand periods, AI-driven optimization contributes to a more stable and sustainable energy infrastructure (ibid).

Use of AI for demand response methods rely on large datasets of real time data from sources, such as smart meters and charging stations for accurate predictions and reliable system operation. By leveraging real-time data, AI-driven dynamic charging schedule optimization enhances grid efficiency while reducing operational costs (Shern 2024). AI is employed in demand forecasting, utilizing historical data and user behavior patterns to predict peak billing periods. This predictive capability enables more effective electricity distribution, reducing the risk of blackouts and ensuring grid stability during periods of high demand. AI also plays a crucial role in dynamic pricing and energy management, where algorithms adjust charging rates in real time based on market conditions, grid load, and energy availability. This adaptive approach enhances grid reliability while optimizing cost-effectiveness for both consumers and energy providers.

2.5 Autonomous driving

Commercially available autonomous vehicles (AVs) are highly likely to be EVs for three primary reasons. First, emission-free and quieter vehicles are more suitable for adoption in densely populated urban areas and align with global efforts to reduce carbon emissions (Nunno 2021). Second, AVs require a substantial and consistent power supply for their sensors and computational systems, which batteries can provide more effectively than internal combustion engines (ibid). Third, EVs exhibit greater responsiveness, enhancing safety and facilitating more efficient control by artificial intelligence (AI) systems (ibid).

Greenblatt and Shaheen (2015) examined the greenhouse gas (GHG) reduction potential of driverless taxis in the United States, asserting that the deployment of such vehicles could lead to an 87–94% reduction in emissions per vehicle-kilometer traveled by 2030. The authors further argued that, compared to conventional fuel-powered and hybrid electric vehicles, driverless taxis could achieve a 63–82% reduction in GHG emissions within the same timeframe (Greenblat and Shaheen 2015). This reduction is primarily attributed to three factors: increased vehicle utilization per year, enhanced fuel efficiency due to the adoption of lighter and more aerodynamically optimized vehicle designs, and decreased GHG emissions resulting from electricity consumption. Vehicle electrification is projected to lower vehicle emission intensities by approximately 11% and reduce regional GHG emissions by over 5% (Le Hiong et al. 2021). Additionally, Hong and Zimmerman estimated that AVs could decrease GHG emissions by 20% compared to a scenario without AVs by the year 2040. Even under a worst-case scenario—where automation induces increased personal vehicle use alongside an 85% electrified vehicle fleet—GHG emissions would still decline significantly (ibid).

Driverless vehicles rely on highly advanced systems, including high-performance computing and an increasing array of ADAS sensors such as high-resolution cameras, radar, and LIDAR (TE Connectivity 2018). They incorporate human-machine interfaces like 4K/8K displays and head-up displays. Fully autonomous (Level 5) vehicles are expected to transmit approximately 25 gigabytes of data to the cloud per hour (kdespagniqz 2015). Additionally, vehicle-to-everything (V2X) communication combines sensor-based and radio-based technologies, enabling AVs to interact with their surroundings and exchange data with other vehicles and traffic infrastructure. However, safe autonomous operation requires real-time data transmission, which current LTE networks, with a latency of 30–40 milliseconds, cannot support (TE Connectivity 2018). To address this, 5G mobile communication is essential for AVs.

Automotive manufacturers are also exploring fiber optic connectivity to mitigate electromagnetic interference. Unlike traditional electronic communication, fiber optics do not suffer from electromagnetic emissions, crosstalk, or signal distortion, allowing them to function reliably near high-power electrical systems without safety concerns (ibid).

3. Discussion and Results

Application of AI in the field of mobility is dependent on the availability and accessibility of a robust digital infrastructure that allows real-time communication and data analysis, compute infrastructure, low-latency communication networks (such as 5G), EV data sources, smartphone availability for shared mobility apps and road conditions that allow ADAS and autonomous mobility to operate successfully. These conditions are easier to meet in the nations of the Global North – referring to the economically developed, industrialized nations, primarily in North America, Europe, and parts of East Asia, distinguished by advanced economies and higher standards of living. However, several countries in the Global South – economically and socially less developed countries, primarily located in Africa, Asia, and Latin America, often characterized by lower income levels, limited industrialization, and challenges in healthcare, education, and infrastructure – lack these conditions prerequisite for enabling AI approaches for smart and sustainable mobility.

Despite the transformative potential of AI, several challenges persist, particularly in the Global South, where infrastructural limitations may impede AI development. The substantial data requirements for training AI systems and the extensive computational resources necessary for this process pose significant barriers in regions with underdeveloped digital infrastructure (Okolo 2023). One of the most pressing challenges in Africa is limited internet access. Over the past decade, internet penetration in Africa has increased considerably, rising from 8% in 2011 to 36% in 2021 (ibid). However, this growth remains constrained by inadequate access to electricity and insufficient investments in essential internet infrastructure, including fiber optic networks, cellular towers, and base stations. According to World Bank data, 80.7% of the urban population in Sub-Saharan Africa has access to electricity, a figure that contrasts sharply with the nearly universal urban electricity connectivity rates of 99.9% in South Asia and 99.5% in Latin America and the Caribbean (World Bank 2025). The disparity is even more pronounced in rural areas, where only 30.4% of the rural population in Sub-Saharan Africa has access to electricity, compared to 98.3% in rural South Asia and 96.5% in rural Latin America and the Caribbean (ibid). These infrastructural deficits present critical obstacles to AI adoption and development, further exacerbating the digital divide between high-income and low-income regions.

Data availability and compatibility present significant challenges in resource-constrained regions, largely due to underdeveloped infrastructure and limited practitioner capacity. The utilization of real-time data for computations based on foundation models is particularly hindered by unreliable network connectivity, which remains a critical barrier to AI implementation (Yu et al. 2023). The successful deployment of modern AI systems is also contingent upon the availability of skilled technical talent. Expertise is required to establish and maintain technical infrastructure, leverage existing AI models and tools, curate and refine datasets, and develop new AI-driven solutions (ibid). Additionally, industry practitioners must undergo continuous upskilling to effectively adopt and integrate AI technologies into their workflows. Compounding these challenges, AI introduces substantial risks that necessitate robust policy frameworks and governance mechanisms. Regulations and guidelines are essential to ensure the responsible and ethical deployment of AI while mitigating potential harms (ibid). Populations in the Global South generally have lower levels of literacy on issues such as data privacy and algorithmic bias, making governance frameworks crucial for guiding AI adoption and safeguarding its societal impact. It is important to acknowledge that many regions in the Global South often lack the extensive resources necessary for the independent development of advanced AI systems. As a result, these countries rely heavily on AI software produced by technologically advanced nations in the Global North (Effoduh 2024). This dependency positions African nations primarily as consumers of AI technologies, which are developed in contexts that may not align with their unique cultural, ethical, and social frameworks. Moreover, Western nations, recognizing the significant market potential in Africa, actively supply AI technologies, thereby fostering a dynamic that reinforces technological dependence (ibid).

Autonomous vehicles and semi-autonomous vehicles depend on minor infrastructure upgrades to enhance lane visibility and standardize marking and maps for AI to correctly steer the vehicle on road (Souweidane and Smith 2023). Unclear and non-standard markings and irregular construction zone variabilities also affect the efficacy of ADAS technologies. For example, ADAS implementation in India faces several challenges due to inconsistent lane markings, diverse vehicle types, and unpredictable traffic conditions, leading to sudden braking or stops. The presence of bikes, rickshaws, stray animals, and pedestrians further complicates detection, as ADAS struggles to interpret their size and movement (Geomechanic 2024). Designed for structured traffic environments, ADAS finds it difficult to adapt to India’s chaotic, nonlinear road systems. Additionally, India lacks advanced semiconductor capabilities, relying on imports from Taiwan, the US, and Singapore for essential ADAS components, making technological advancement in this sector even more challenging (Singh 2024).

Appreciating these overarching challenges with regards to general adoption of AI in Global South countries, a focused analysis of the mobility sector’s challenges to AI adoption is warranted. Understanding the challenges in a few Global South countries can aptly highlight the impediments to implementation of AI in the mobility sector. For this study, the availability of necessary infrastructure in ten Global South countries with the highest vehicular population is considered and analyzed, as implementing AI for sustainable mobility in these countries can significantly influence climate action against climate poly-crisis. The parameters considered in the analysis are as follows:

1. Vehicular population (P1): Number of cars (in million) in a country in 2025 (World Population Review 2025a)
2. Internet penetration (P2): Individuals using the internet as percentage of the country’s population (World Bank 2025)
3. 5G penetration (P3): Density of 5G deployment in the country (nPerf 2025)
4. Cellphone penetration (P4): Number of cell phones in use per 100 people in the country (World Population Review 2025b)
5. Traffic layer data quality in Google Maps (P5): Quality of traffic level data to users and developers for route planning (Google 2025)

Table 1 shows the countries, with the highest vehicular population in the Global South, that have been assessed.

Table 1: Assessment of AI enablement factors in 11 countries of Global South with highest vehicular population.

From the aforementioned secondary data available in public domain, it is evident that countries in the Global South have varied level of infrastructural maturity for implementation of AI in sustainable mobility. For example, in China, while the vehicular population is highest among all Global South countries, the penetration of 5G network is limited to the urban areas, while the other regions lack 5G connectivity, which can limit capabilities in real-time route planning, demand prediction for EV charging network and autonomous cars. In contrast, India has a robust 5G network across the nation, however, the lack of cellphone and internet penetration limits its potential to tap into AI’s capabilities to deploy AI for smart and sustainable mobility equitably across the nation. Though availability of data from Google Maps API is consistently good across most of the nations, the lack of one or more enabling factors limits the capabilities of the Global South nations to fully realize the benefits of AI for tackling the climate poly-crisis in automotive industry.

Owing to the inaccessibility of digital infrastructure and usability challenges associated with platforms on which route optimization methods depend, AI emerges as an unavailable technology in Global South for sustainable mobility. Moreover, unaffordability of modern cellphones for the low-income households and inapplicability of ethical use of ADAS technologies in the congested roads of the Global South posit AI as an unavailable technology for the population of these regions.

The author argues that in the current landscape of inequitable availability of necessary infrastructure for AI-powered sustainable mobility, AI does not present itself as an immediate remedy for tackling the climate poly-crisis that the automotive sector is contributing to. AI can be posited as a technological solution to climate-hostile mobility systems in the Global North, as they have the required infrastructure for enablement of AI solutions; however, for Global South countries to leapfrog in sustainable mobility space using AI, considerable investments and technology transfers are required. An analysis of how long AI deployment can be done in the mobility space of Global South countries and if the time period is sufficiently short for tackling climate crisis needs to be done; however, has been kept outside the scope of this review paper.

4. Future directions and Conclusion

Although the review paper assesses the applicability of AI for smart and sustainable mobility solutions in the Global South countries, further research is required to understand how and by when these countries can leapfrog to leverage AI for sustainable and safe mobility. Policy considerations have been kept outside the scope of this paper, however, recommendations for policies suited to the economic, political and social landscape of the Global South countries need to be explored. Analysis of consumer attitude towards AI solutions in mobility and acceptance or adoption trends of AI need to be examined as well.

AI serves as a futuristic technology for enabling new mobility solutions in the Global South countries and can reduce the climate footprint of the automotive sector in these regions. However, given the lack of the fundamental technological infrastructure for AI deployment limits the potential of AI as an enabling agent of sustainable and safe mobility in the Global South countries. Other solutions based on existing technologies and digital infrastructure need to be explored to facilitate adoption of climate-friendly mobility solutions in the Global South countries.

References

1. 5G map: Cellular 5G networks in the world (2025). nPerf. https://www.nperf.com/en/map/5g. Accessed 31 Mar 2025
2. AI for Smart Cities: Building the Infrastructure for the Future (2025). BP3 Global. https://www.bp-3.com/blog/ai-powered-smart-cities-building-tomorrows-infrastructure
3. Amzolin G, Haraguchi N, De Sousa APN, Savrasov A, Reis J (2024). Bridging the AI Divide: Empowering Developing Countries through Manufacturing. United Nations Industrial Development Organization. https://www.unido.org/sites/default/files/unido-publications/2024-10/IID%20Policy%20Brief%2012_0.pdf. Accessed 22 Mar 2025
4. Bahamazava K (2025) AI-driven scenarios for urban mobility: Quantifying the role of ODE models and scenario planning in reducing traffic congestion. Transport Economics and Management. https://doi.org/10.1016/j.team.2025.02.002
5. Balakumar P, Ramu SK, Vinopraba T (2024) Optimizing electric vehicle charging in distribution networks: A dynamic pricing approach using internet of things and Bi-directional LSTM model. Energy, 294. https://doi.org/10.1016/j.energy.2024.130815
6. Bulusu A (2023) What Companies Should Know About Route Optimization Technology. Forbes. https://www.forbes.com/councils/forbestechcouncil/2023/06/02/what-companies-should-know-about-route-optimization-technology/ Accessed 22 Mar 2025
7. Carlin D, Arshad M (2024) Climate Risks in the Transportation Sector. United Nations Environment Programme Finance Initiative. https://www.unepfi.org/wordpress/wp-content/uploads/2024/05/Climate-Risks-in-the-Transportation-Sector.pdf. Accessed 15 Mar 2025
8. Cars by country (2025). World Population Review. https://worldpopulationreview.com/country-rankings/cars-by-country. Accessed 31 Mar 2025
9. Cell Phones by Country (2025). World Population Review. https://worldpopulationreview.com/country-rankings/cell-phones-by-country. Accessed 31 Mar 2025
10. Chung Y, Cho H, Choi K (2013) Impacts of freeway accidents on CO2 emissions: A case study for Orange County, California, US. Transportation Research Part D: Transport and Environment, 24, 120-126. https://doi.org/10.1016/j.trd.2013.06.005
11. Costa CJ, Aparicio M, Aparacio S, Aparacio JT (2024) The Democratization of Artificial Intelligence: Theoretical Framework. Applied Science 14(18), 8236. https://doi.org/10.3390/app14188236
12. Effoduh J (2024) A Global South Perspective on Explainable AI. Carnegie Endowment for International Peace. https://carnegieendowment.org/research/2024/04/a-global-south-perspective-on-explainable-ai. Accessed 30 Mar 2025
13. European Commission (2024) Road safety thematic report – Main factors causing fatal crashes. European Road Safety Observatory. Brussels, European Commission, Directorate General for Transport.
14. European Environment Agency (2019) Transport: increasing oil consumption and greenhouse gas emissions hamper EU progress towards environment and climate objectives. https://www.eea.europa.eu/en/analysis/publications/transport-increasing-oil-consumption-and. Accessed 15 Mar 2025
15. EV Industry Summit (2023) 6 Ways AI Will Change EV Charging. https://evchargingsummit.com/blog/how-ai-will-change-ev-charging. Accessed 30 Mar 2025
16. Fagnant DJ, Kockelman K 2015) Preparing a nation for autonomous vehicles: opportunities, barriers and policy recommendations. Transportation Research Part A: Policy and Practice, 77, 167-181. https://doi.org/10.1016/j.tra.2015.04.003
17. Geduld H, Sinclair M, Steyn E, Chu K (2024) Road Traffic Injuries: Solutions to a complex problem. Annals of Global Health. https://annalsofglobalhealth.org/articles/10.5334/aogh.4249. Accessed 15 Mar 2025
18. Geomechanic (2024) Why does ADAS Stand Unsuccessful on Indian Roads? https://gomechanic.in/blog/why-does-adas-stand-unsuccessful-on-indian-roads. Accessed 31 Mar 2025
19. Google Maps Platform Coverage Details (2025). Google. https://developers.google.com/maps/coverage. Accessed 31 Mar 2025
20. Greenblatt JB, Shaheen S (2015) Automated Vehicles, On-Demand Mobility, and Environmental Impacts. Current Sustainable/Renewable Energy Reports,2, 74–81. https://doi.org/10.1007/s40518-015-0038-5
21. Huether P (2024) Electric Vehicle Efficiency: Unlocking Consumer Savings and Environmental Gains. American Council for an Energy-Efficient Economy. https://www.aceee.org/sites/default/files/pdfs/electric_vehicle_efficiency_-_unlocking_consumer_savings_and_environmental_gains.pdf. Accessed 29 Mar 2025
22. IMF (2024) AI Preparedness Index. International Monetary Fund. https://www.imf.org/external/datamapper/datasets/AIPI. Accessed 22 Mar 2025
23. International Telecommunication Union (ITU) World Telecommunication (2025) Individuals using the Internet (% of population). World Bank. https://data.worldbank.org/indicator/IT.NET.USER.ZS. Accessed 31 Mar 2025
24. ITF (2019) ITF Transport Outlook 2019. OECD Publishing. https://doi.org/10.1787/transp_outlook-en-2019-en. Accessed 15 Mar 2025
25. ITSA (2023) The Impact of AI on Transportation and Mobility. Intelligent Transportation Society of America and Cambridge Consultants. https://itsa.org/wp-content/uploads/2023/12/Lit-R-018-v0.3-2023-ITSA-AI-report.pdf. Accessed 15 Mar 2025
26. Karimipour H, Tam VWY, Le KN, Burnie H (2021) Routing on-road heavy vehicles for alleviating greenhouse gas emissions. Cleaner Engineering and Technology, 5. https://doi.org/10.1016/j.clet.2021.100325
27. kdespagniqz (2015) Connected cars will send 25 gigabytes of data to the cloud every hour. Quartz. https://qz.com/344466/connected-cars-will-send-25-gigabytes-of-data-to-the-cloud-every-hour. Accessed 30 Mar 2025
28. Khan MS, Umer H, Faruqe F (2024) Artificial intelligence for low income countries. Humanities and Social Sciences Communications 11, 1422. https://doi.org/10.1057/s41599-024-03947-w
29. Le Hong Z, Zimmerman N (2021) Air Quality and Greenhouse Gas Implications of Autonomous Vehicles in Vancouver, Canada. Transportation Research Part D: Transport and Environment,90. https://doi.org/10.1016/j.trd.2020.102676
30. Lungu MA (2024) Smart Urban Mobility: The Role of AI in Alleviating Traffic Congestion. Proceedings of the International Conference on Business Excellence 18(1):1441-1452. https://doi.org/10.2478/picbe-2024-0118
31. Masello L, Castignani G, Sheehan B Murphy F, McDonnell K (2022) On the road safety benefits of advanced driver assistance systems in different driving contexts. Transportation Research Interdisciplinary Perspectives, 15. https://doi.org/10.1016/j.trip.2022.100670
32. Mohsen B (2024) AI-Driven Optimization of Urban Logistics in Smart Cities: Integrating Autonomous Vehicles and IoT for Efficient Delivery Systems. Sustainability 2024, 16(24), 11265; https://doi.org/10.3390/su162411265
33. Mohsen, B. (2023) Impact of Artificial Intelligence on Supply Chain Management Performance. Journal of Service Science and Management, 16, 44–58. https://doi.org/10.4236/jssm.2023.161004
34. Mohsen, B.M. (2023) Developments of Digital Technologies Related to Supply Chain Management. Procedia Computer Science, 220, 788–795. https://doi.org/10.1016/j.procs.2023.03.105
35. Nagpal RK, Cohen E (2022) Automotive electronics revolution requires faster, smarter interfaces. Embedded. https://www.embedded.com/automotive-electronics-revolution-requires-faster-smarter-interfaces. Accessed 29 Mar 2025
36. Narasimhan T (2024) Key Trends in Global Advanced Driver Assistance Systems (ADAS) and Autonomous Driving (AD) Market. Frost & Sullivan. https://www.frost.com/growth-opportunity-news/key-trends-in-global-advanced-driver-assistance-systems-adas-and-autonomous-driving-ad-market. Accessed 29 Mar 2025
37. Numalis (2025) Driving into the Future: The Unstoppable Rise of AI in Transportation. https://numalis.com/ai-in-transportation. Accessed 29 Mar 2025
38. Nunno R (2021) Autonomous Vehicles: State of the Technology and Potential Role as a Climate Solution. Environmental and Energy Study Institute. https://www.eesi.org/files/IssueBrief_Autonomous_Vehicles_2021.pdf. Accessed 30 Mar 2025
39. Okolo CT (2023) AI in the Global South: Opportunities and challenges towards more inclusive governance. Brookings. https://www.brookings.edu/articles/ai-in-the-global-south-opportunities-and-challenges-towards-more-inclusive-governance. Accessed 30 Mar 2025
40. Pacheco-Capitaine HMGM (2023) Artificial intelligence and advanced driver assistance systems absorption (ADAS) in Mexico. Ciencia Nicolaita, 88. https://doi.org/10.35830/cn.vi88.614
41. Santilli P (2023) Singapore Best Practices in Intelligence Smart City Applications. Strategic Consortium for Intelligence Professionals. Available online: https://www.scip.org/news/647788/Singapore-Best-Practices-in-Intelligence-Smart-City-Applications.htm, accessed on 24 March 2025
42. Schellekens P, Skilling D (2024) Three Reasons Why AI May Widen Global Inequality. Center for Global Development. https://www.cgdev.org/blog/three-reasons-why-ai-may-widen-global-inequality. Accessed 22 Mar 2025
43. Shaheen S, Cohen A, Zohdy I (2016) Shared mobility: Current practices and guiding principles (FHWA-HOP-16-022). US Department of Transportation. https://ops.fhwa.dot.gov/publications/fhwahop16022/fhwahop16022.pdf
44. Shern SJ, Sarker MT, Ramasamy G, Thiagarajah SP, Farid FA, Suganthi ST (2024) Artificial Intelligence-Based Electric Vehicle Smart Charging System in Malaysia. World Electric Vehicle Journal, 15(10), 440; https://doi.org/10.3390/wevj15100440
45. Singh MY (2024) The Core Issue With ADAS Is Not Just Automotive Technol­ogy But An AI Challenge. Electronics for You. https://www.electronicsforu.com/technology-trends/core-issue-with-adas-not-just-automotive-technology-but-an-ai-challenge. Accessed 31 Mar 2025
46. Souweidane N, Smith B (2023) State of ADAS, Automation, and Connectivity. Center for Automotive Research. https://www.cargroup.org/wp-content/uploads/2023/03/State-of-ADAS-Automation-and-Connectivity-Report.pdf. Accessed 31 Mar 2025
47. TE Connectivity (2018) 6 Key Connectivity Requirements of Autonomous Driving. IEEE Spectrum. https://spectrum.ieee.org/6-key-connectivity-requirements-of-autonomous-driving. Accessed 30 Mar 2025
48. Theissler A, Pérez-Velázquez J, Kettelgerdes M, Elger G (2021) Predictive maintenance enabled by machine learning: Use cases and challenges in the automotive industry. Reliability Engineering & System Safety, 215. https://doi.org/10.1016/j.ress.2021.107864
49. Tikoudis I, Martinez L, Farrow K, Bouyssou CG, Petrik O, Oueslati W (2021) Exploring the impact of shared mobility services on CO2; OECD Environment Working Papers No. 175. Organisation for Economic Co-operation and Development. https://dx.doi.org/10.1787/9d20da6c-en
50. Trotolo F, Ahmed A, Hayat H, Hayat D (2024) What Are the Trends of AI Within the Transportation Industry? Walturn. https://www.walturn.com/insights/what-are-the-trends-of-ai-within-the-transportation-industry. Accessed 29 Mar 2025
51. Tunnell J, Asher ZD, Pasricha S, Bradley TH (2018) Toward Improving Vehicle Fuel Economy with ADAS. SAE International Journal of Connected and Automated Vehicles, 1(2). https://doi.org/10.4271/12-01-02-0005
52. van Brecht J, van Gent, MJ, Renes RJ (2022) CO2 reduction through shared mobility: A review of psychological factors for the switch from private car to shared transport. Research group Psychology for Sustainable Cities – Amsterdam University of Applied Sciences.
53. Vélez AMA (2023) Environmental impacts of shared mobility: a systematic literature review of life-cycle assessments focusing on car sharing, carpooling, bikesharing, scooters and moped sharing. Transport Reviews, 44(3), 634–658. https://doi.org/10.1080/01441647.2023.2259104
54. Vulog (2025) AI Damage Management and Detection. https://www.vulog.com/blog/ai-in-shared-mobility-damage-detection-and-management. Accessed 29 Mar 2025
55. Wang Y, Lv C, Nie Q, Liu H (2024) Analyzing the Impact of Road Accidents on Carbon Dioxide Emissions in Freeway Traffic: A Simulation and Statistical Modeling Approach. Sustainability 2024, 16(5), 2168. https://doi.org/10.3390/su16052168
56. Wood I (2024) New EU safety regulations mandate the use of ADAS. Autonomous Vehicle International. https://www.autonomousvehicleinternational.com/news/legislation/new-eu-safety-regulations-mandate-the-use-of-adas.html. Accessed 29 Mar 2025
57. World Bank (2025) Access to electricity, urban (% of urban population). https://data.worldbank.org/indicator/EG.ELC.ACCS.UR.ZS. Accessed 30 Mar 2025
58. Yu D, Rosenfeld H, Gupta A (2023) The ‘AI divide’ between the Global North and the Global South. World Economic Forum. https://www.weforum.org/stories/2023/01/davos23-ai-divide-global-north-global-south. Accessed 30 Mar 2025
59. Zhang, Y., Wang, Y., & Li, X. (2021) Reinforcement Learning-Based Traffic Signal Control: A Comprehensive Review. IEEE Transactions on Intelligent Transportation Systems, 22, 3771-3787.