Special collections for applying artificial intelligence techniques to encourage economic growth and maintain sustainable societies

Abstract

Efficient resource use requires substantial evaluation and re-evaluation of production processes. Since not all production details can be properly and timely monitored and adjusted, improvement of resource allocation has long been a critical issue in commodity production, technique selection, and sustainable development. The progress of artificial intelligence (AI) offers a possibility. For example, with deep learning techniques and extensive data analysis, most previously unincorporated or unknown information associated with a firm's activities can be appropriately reflected and calculated. Once the firms take the report, a more efficient and economically friendly production strategy could be made. The central theme of this special collection is to invite studies on how the design and application of AI benefit not only the fields of computer science and information engineering but also the interdisciplinary fields, including renewable energy development, environmental protection, and economic analysis. Fourteen papers are published in this special collection.

Introduction

Pursuing sustainable social, economic, and environmental development is critical to improving intergeneration equity. However, it is not generally achievable due to inefficient resource allocation in production processes. Since production strategies are highly interrelated and interdependent with social behaviors, sustainable development must also balance various societal and environmental decisions; otherwise, resources of any kind would not be optimally allocated. For example, profit-maximizing private sectors tend to overuse resources and cause externalities because they equate their private marginal benefits and private marginal costs associated with the production. In other words, the firms would not, intentionally or unintentionally, internalize the social costs (i.e. environmental degradation, water pollution, etc.) imposed on the public. As a result, the resources (land, capital, water, etc.) are generally not allocated efficiently, making social optimum unattainable.

While finding solutions to the problem of societal sustainability is hard to achieve due to their inherent complexity, the recent progress of artificial intelligence (AI) makes optimal resource allocation feasible and applicable because more local, regional, or national characteristics can be appropriately incorporated and evaluated. For example, AI can help determine the distribution of charging stations for hybrid and electric vehicles, the selection of production strategies in the face of short-term climate events or long-term climate impacts, and environmental conditions with timely monitoring and evaluation systems.

Launching this special is to welcome insightful works addressing these issues. It thus could be beneficial to the decision-makers, researchers, and the general public who formulate policies (i.e. to regulate production), determine investment decisions (i.e. project managers), and care about the environmental and economic sustainable issues (i.e. pollution level, abatement technology, etc.).

We hope that with the completion of this special collection, more researchers can realize that the design, advances, and utilization of AI techniques are not only applicable in the fields of computer science and information engineering but also valuable in many interdisciplinary fields, including renewable energy development, environmental protection, and economic analysis. For example, AI can help with optimizing the allocation of charging stations for hybrid and electric vehicles and designing optimal solar and wind power systems, deducing and combining environmental strategies in the face of climate change, and even finding possible solutions for the distribution of social welfare, all of which are worth investigating by young scholars who are interested in energy security, environmental sustainability, biodiversity, and social equity.

Contributions of collections

In the first article, Kung et al. ¹ point out that because biofuel production is highly related to the stability of biomass supply, investigations of biofuel applications’ economic and environmental effects must also incorporate factors such as climate conditions and cropping decisions. The authors propose a theoretical framework that can be used to assess the optimal production of agricultural commodities and biofuel production under climate change and then simulate the model, using Taiwan as an example, to address how such a framework could be helpful in policy analysis. The authors indicate that a goal to optimize biofuel production and emission reduction may not always be beneficial if the loss of agricultural commodities or a shift in cropping activities is substantial. They also point out that using AI monitoring systems could collect and analyze agricultural data so that the firms could promptly adjust their production strategies and subsequently improve efficiency.

The second paper by Cao et al. ² employs the InVEST and FLUS models and uses 30-year Qiantang River basin data to evaluate the impacts of land-use change on carbon storage. The authors find that total carbon storage rapidly declined from 2000 to 2020. To make the results robust, they simulate a base scenario based on the historical trend development, then evaluate the results with food security and ecological protection scenarios. For the food security scenario, the net carbon storage could further decline by 2 million tons in 2030, but that of the environmental protection scenario would increase by 62.60 tons per km². The authors also highlight how AI methods may be incorporated in subsequent studies. This article thus could be a valuable reference for decision-makers to formulate carbon storage and land management policies under the scope of climate change and sustainable development.

In the following paper, Kim and Cho ³ use real-time lending data to predict the ability of borrowers to repay the peer-to-peer (P2P) debt. The authors point out that since most debt is not approaching maturity date and the borrowers are not obligated to repay this debt, it is necessary to propose divertive convolutional neural networks (CNNs) under an uncertainty handling scheme to make the prediction more precise. With this formulation, the additional data generated by two semi-supervised learning methods are incorporated by weighted voting. The authors also apply a diversity measure derived from various models of CNNs to extract the data details, such as the labeling noise and the loan status. The results show that combining the actual dataset and the advanced weighted voting mechanism would outperform the unweighted CNNs methods.

The fourth study by Zhang et al. ⁴ indicates that while solar energy development would improve energy security and reduce emissions, the climate-induced impacts on sunny hours and solar radiation must be further investigated. This article uses the 2009 to 2021 climate data of Jiangxi province to evaluate the changes in regional sunny hours and the strength of solar radiation. The authors further employ time series models (i.e. ARIMA and ARDL) to predict the future changes in climate factors from 2021 to 2025. The results show geographical differences in solar energy generation and validate that solar energy potential must incorporate regional characteristics. The authors point out that, when considering climate factors, the power potential from the alternative solar panel systems (i.e. conventional PV, PV/PCM, and PVT/PCM) would decrease by approximately 200 kWh/m² per year. Given the same investment and installation, Taiwan's total solar power potential may decline by up to 8%. Therefore, the Taiwanese government must react to the climate impacts to minimize the power supply fluctuation and ensure effective solar energy development.

The central theme of the fifth paper by Wang et al. ⁵ uses 2010–2020 data to investigate the “Environment–Resource–Energy–Economy” nexus along the Yangtze River Delta. Specifically, the authors develop a non-linear input–output system that incorporates carbon emission and other pollutions to evaluate the level of green industrial development. They find that the agricultural total factor productivity (AGTFP) has gradually increased, partly due to the increased density in economic agglomeration. However, economic agglomeration would have a multiple-threshold effect: a weak facilitative effect in the early stage, followed by land fragmentation in the growth stage and an increasing return to scale in the maturity stage. The authors thus claim that China should further promote economic agglomeration because it would benefit the AGTFP and contribute to both carbon sequestration and food security.

The sixth paper by Lin et al. ⁶ analyzes how residents could participate in urban renewal when there is no trust in the builder and indeterminate benefits. In recent years, the completion rate of urban renewal has been less than 0.1%. Thus, the authors employ the theory of planned behavior to investigate what factors may influence this situation the most and help formulate renewal policies in Taiwan. Based on the 545 valid questionnaires, the results show that if the residents are unaware of their perceived benefits from the renewal projects, they will likely have a low participation rate. The authors further identify that the key to increasing the willingness of renewal participation is that the government can force the builders to complete the projects on time completion in advance.

In the following paper, Yu et al. ⁷ incorporate the International Federation of Robotics and air pollution data from 2013 to 2018 to investigate whether industrial robots can be used as a primary intelligent manufacturing technology to reduce air pollution in China. They find that, even with the possible existence of endogeneity, industrial robots could significantly improve air quality in terms of PM2.5, PM10, and SO₂ levels. The probable explanation for the results is that the synergistic benefits of employing industrial robots could improve energy efficiency and promote green technology innovations. The authors further use a heterogeneity analysis to show that industrial robots should be considered a means of green technology because it benefits areas focused on green policies such as low-carbon piloting, environmental regulation, and resource planning. Furthermore, the authors claim that such an application would help control pollution and improve industrialization.

The central theme of the paper by Chen et al. ⁸ is to summarize how incorporating machine learning (ML) approaches would increase affect biochar production and to the extent it increases economic and environmental benefits from biochar applications. This article first provides the background of the biochar production process and discusses the approaches that can be used to remove heavy metals and other contaminations. Then, it illustrates the possible effects on food security, energy security, and carbon sequestration associated with the biochar application. The authors further prove that the ML can improve the prediction accuracy of soil and water quality adjacent to the land applied with biochar by approximately 12.7%. This article suggests that ML approaches can significantly reduce costs associated with experimental trials and working hours. However, whether such a reduction can be considered optimum requires further investigations.

The next paper by Liao et al. ⁹ incorporates the concept of financial compensation to evaluate the effects of the natural forest protection system on China's ecological conservation and socio-economic development policies. The contingent valuation method is applied based on the survey data from 486 farming households. The results indicate that 72% of farmers are willing to participate in the protection system at a per-hectare compensation of ￥517.95 per year, significantly higher than the current subsidy level of ￥225. This result partially explains the failure of the existing policy, and the authors further raise some suggestions that the government could react to achieve sustainable forestry.

The tenth paper by Hou ¹⁰ builds an enriched Environmental Kuznets Curve framework to evaluate how changes in hazardous industrial discharge affect economic growth and green innovations from 2007 to 2021. The results indicate that the pollution level is significantly related to the total output, and for every 1% increase in production, approximately 444.6 tons of SO₂ would be emitted. The authors also point out that the rapid rise in wastewater in the regions that are relatively less developed is primarily caused by the economic growth in the last decade. While these effects in large, better-developed cities are undetermined, the results demonstrate that more strict environmental policies have considerably achieved green production transformation.

The main theme of the eleventh paper by Zhang et al. ¹¹ deals with the issue of horizontal logistics collaboration and effective supply chain under the approach of coding analysis. The authors construct a theoretical framework of a horizontal logistics coalition of 41 primary Chinese logistics service providers to investigate the impacting factors of their service and competition power. The results indicate that the providers’ thoughts, the market situation's reaction, and the strategy's adjustment are the most important factors affecting profitability and market share. The authors conclude that while the driving forces could alter among cities, the effective utilization of logistics workers and the on-time adjustment based on integrating AI technology would be beneficial.

In the following paper, Wu and Zuo ¹² consider whether AI techniques provide more possibilities for supply chain transformation to overcome the adverse effects of global warming and environmental degradation. A Cournot game model with an upgrading ML technology is employed to evaluate the use of green technologies of two competing supply chains. The authors point out that the investment risk could be alleviated if the information is symmetric or asymmetric. Furthermore, the market equilibrium of the duopoly model would remain constant if the ML technology is properly upgraded. However, the risk associated with technology upgrades under asymmetric information is vital when a perfectly competitive market is examined. The authors thus conclude that the government should provide more support for firms to employ more ML systems and make the manufacturing process more effective so that carbon emissions could be further reduced.

The thirteenth paper by Sui and Xie ¹³ points out that more contactless services emerge during the problematic period of COVID-19. Thus, the construction of AI on campus would be able to reduce the infection rate of COVID-19. The results show a high correlation between the structure of AI services and the low infection possibility, and to achieve this outcome, campus-wide informatization, a means of a core intermediary mechanism of management that reduces security risks, must be developed appropriately. The authors conclude that the administration should make more efforts to promote the construction of intelligent campuses.

The central theme of the fourteenth paper by Li et al. ¹⁴ focuses on the relationships between the turnover of party secretaries and mayors and the quality of economic development in terms of environmental total factor productivity growth in China. The authors indicate that in order to get a promotion, the leaders would try to decrease the political uncertainty caused by economic development. In other words, the officers would make more efforts on economic rather than environmental development to ensure the turnover could be successful. This result is well explained by the observations of production technology innovations and the number of government interventions.

Conclusion

The papers published in this Special Collection provide novel insights and detailed analyses of how AI methods may be employed in the areas of food security, business administration, renewable energy production, climate change mitigation, and public safety, all of which are vital to social development, economic growth, technology innovation, and environmental sustainability. While these papers employ multiple and complex techniques or approaches, the ultimate reason for this collection is not to establish a complicated model but an attempt to gain insight into universally applicable frameworks that integrate crucial issues of computer science, economics, and environmental science. The discourse is essential to distribute findings, share ideas, stimulate future studies, and motivate the actions of both practitioners and decision-makers.