Can artificial intelligence help accelerate the transition to renewable energy?

The world is shifting away from fossil energy systems toward renewable energy (RE) (e.g., hydropower, solar, and wind) systems (Ahmad et al., 2021; Qin et al., 2023a), aiming to achieve a low-carbon economy (Gyimah, 2022; Su et al., 2023a). Artificial intelligence (AI), a collection of technologies that can imitate intelligent human behavior (Lyu and Liu, 2021; Liu et al., 2021a),¹ is considered to have tremendous potential to support this energy transition (Višković et al., 2022). For example, AI will help predict weather patterns and energy consumption to determine the optimal times and locations for wind energy generation. Moreover, AI is expected to forecast RE generation and calculate the amount of energy stored, allowing companies to adjust their operations accordingly (Abdalla et al., 2021). In recent years, policymakers worldwide have emphasized the importance of AI development. For instance, the US Department of Energy announced that it would invest US$37 million in AI research and development (R&D) in August 2020. Likewise, the UK funds numerous research hubs to develop robotic technology to improve offshore wind safety. In India, AI is expected to add US$50–55 billion to the energy and industry sectors (Chawla et al., 2022).

Although many countries support AI applications in the energy industry, its usage in the RE sector remains limited (Cheng and Yu, 2019). The main obstacle is that the proposed AI methods for optimizing RE are expensive and complex (Jiang and Raza, 2023). Finding a reputed software provider and configuring the software is a time-consuming process (Jimenez and Gonzalez, 2022). Hence, in the short term, AI may not aid the transition to RE. Furthermore, because AI is useful in the RE and non-RE sectors, its development may not necessarily increase the share of RE in total energy; hence, it cannot ensure a rapid energy transition. For example, autonomous robots with AI are used in the management and maintenance of power plants in harsh environments(Chawla et al., 2022; Wong et al., 2018), improving the efficiency of the RE and non-RE sectors. Another example is the widespread use of AI in forecasting fossil fuel generation, demand, and electricity prices (Ahmad et al., 2021; Al-Fattah and Aramco, 2021), benefiting the non-RE sector while slowing the transition to RE in the short term. Therefore, it is unclear whether AI helps accelerate the transition to RE.

This study uses China as a sample to examine the impact of AI on the energy transition over different periods. In recent years, China has accelerated AI development (Liu et al., 2021a; Fatima et al., 2020; Qin et al., 2023b, Qin et al., 2023c). In 2017, the Chinese government issued the “Development Plan for a New Generation of Artificial Intelligence,” aiming to make China a leading AI innovation center by 2030. According to the China Center for Information Industry Development, by 2021, the country had contributed 70.9% of the world’s AI patents, demonstrating China’s growing global influence in the AI industry. Meanwhile, as the world’s largest energy consumer, China is promoting the transition to RE to achieve carbon neutrality (Zhao et al., 2023a, Zhao et al., 2023b). In 2023, the National Energy Administration issued the “Guiding Opinions on Accelerating the Development of Digital and Intelligent Energy,” proposing the use of AI to accelerate the transition to RE. Hence, we can expect AI will aid the transition to RE. However, the effects of AI on energy transition remain inconclusive. As of 2021, coal accounts for more than 50% of China’s total energy production and consumption (Liu et al., 2021b). Hence, in addition to encouraging the use of AI in the RE sector, the government is actively promoting the use of AI in non-RE sectors (e.g., oil, gas, and coal). For example, in 2020, the government issued the “Guiding Opinions on Accelerating the Smart Development of Coal Mines,” advocating intelligent production in coal mines. Due to AI’s importance in non-RE sectors, AI may even slow the transition to RE in the short term. Therefore, the impact of AI on China’s energy transition is uncertain at various points in time and requires further investigation. This paper is of great significance to policymakers in understanding the mechanism of AI’s impact on the energy transition and implement appropriate policies to accelerate the transition.

This paper makes the following contributions. First, previous studies have focused on AI’s applications in energy from a technical perspective (Al-Fattah and Aramco, 2021; Hua et al., 2022), finding that AI will promote RE development. However, they ignore the process and mechanism of AI’s impact on the energy transition. AI will benefit the RE and non-RE sectors, resulting in an uncertain impact on RE’s share. To fill this gap, we investigate the impact of AI on the share of RE production and investment in different periods, helping policymakers understand AI’s impact on the energy transition and develop appropriate policies. Second, the study uses the wavelet-based quantile-on-quantile (QQ) method, enabling the examination of different quantiles of AI on different quantiles of the energy transition. Jimenez and Gonzalez (2022) investigated the use of AI in Latin America’s energy transition but did not conduct a quantitative analysis. To fill this gap, we use the wavelet-based QQ method. This method combines wavelet analysis and the QQ approach. We use wavelet analysis to extract a smooth trend from raw data at different time scales. Then, using the QQ approach, we examine the impact of different AI quantiles on different RE quantiles. The results show that AI does not significantly promote the share of RE at the lower quantiles of AI. However, AI increases the share of RE at the upper AI quantiles, implying that the maturity of AI will significantly promote the development of RE. Furthermore, the results vary across periods. AI reduces RE’s share in the short term, primarily because the non-RE sector benefits more from AI development than the RE sector. However, in the long term, AI will eventually help China’s transition to RE.

The remaining sections of the paper are divided as follows. The second section reviews the literature. The third section provides a theoretical analysis of AI’s impact on the energy transition. The fourth section presents the empirical method. The fifth section describes the data. The sixth section presents the empirical results, which are discussed in detail in the seventh section. The final section discusses the conclusion and implications of the findings.