Leveraging AI for Decarbonization in the Energy Sector: Opportunities, Challenges, and Considerations

February 17, 2026

Reading time: 24 minutes

Preface

Please note: this is a reproduction of an article from a special report on AI for Energy, Water, and Waste Management: Unlocking Opportunities, Navigating Challenges, and Learning from Experience co-produced by the Veolia Institute and Microsoft. Contributing authors hold copyright to their article. All materials may be copied and used provided it is properly cited.

Citation: Nilushi Kumarasinghe, Ursula Eicker, Shahin Masoumi-Verki, Kathryn Kaspar, Suchit Ahuja, Damon Matthews. (2026). Leveraging AI for Decarbonization in the Energy Sector: Opportunities, Challenges, and Considerations. Microsoft Corporation and the Veolia Institute. https://www.institut.veolia.org/en/leveraging-ai-decarbonization-energy-sector

Authors

Publisher

Co-produced by the Veolia Institute and Microsoft.

Abstract

Decarbonizing the energy sector while meeting increasing energy demand is one of the central challenges in the global response to climate change. This article discusses the ways in which artificial intelligence (AI) can help to decarbonize the energy sector, and the risks and challenges that can arise from AI-led innovation. We provide evidence-based examples of AI applications in energy supply, generation, and consumption that support decarbonization pathways. Additionally, future potential trajectories for AI in the energy sector are explored and discussed. Challenges to the uptake and scalability of these AI solutions include existing societal inequalities, financial and infrastructure gaps, and the uncertainty around rebound effects of AI. While AI technologies show much promise, realizing their potential to help achieve decarbonization targets will require addressing current inequalities and unsustainable political, societal, and economic behaviors.

Introduction

The global energy sector is central to climate change response strategies around the world. The sector remains the largest contributor to carbon dioxide (CO₂) emissions (two-thirds of global emissions)¹ and the second largest anthropogenic source of global methane (CH₄) emissions,² making this sector the dominant driver of climate change. Although renewable energy supply has been growing rapidly and has begun to curb the energy sector’s climate footprint, energy supply derived from fossil fuels (coal, oil, and gas) continues to grow, albeit at a decreasing rate. In turn, the energy sector itself is being impacted by climate change (damage to infrastructure from extreme weather events) and demand for electricity is rising due to several factors, including population growth, economic development, and the electrification of the building and transportation sectors.¹ The energy sector must meet this growing demand for energy, while also dramatically reducing its emissions within the coming decade.

This article explores how AI can play a role in addressing these key challenges and potentially accelerate the energy sector’s transition towards a net carbon zero pathway. We identify opportunities for AI-supported interventions and provide details of existing applications and success stories. The article also explores risks and challenges, as they relate to the energy sector, which can arise from AI-led innovation and provides important reflections for policy makers and other stakeholders within the energy sector.

AI and the Energy Sector

The energy sector can be subdivided into three key areas (see Figure 1). Energy and mineral supply refers to the extraction of energy sources such as oil and natural gas, as well as other raw materials such as critical minerals needed for renewable energy production. Energy transformation or generation includes the processes needed to transform the energy source to its end use form, such as burning fossil fuels to generate electricity or refining crude oil for fuel, as well as the processes needed to transport this end product to users. Energy consumption includes the various ways in which it is used by consumers or industries, such as heating buildings, driving vehicles or manufacturing in factories.³ Each of these processes emit CO₂ and AI can and is already being used in multitude ways to help reduce emissions. The ability of AI to quickly learn from datasets and identify patterns can strengthen decision making processes and enhance efficiencies across complex systems that exist throughout the energy sector.⁴

AI and Energy Supply

Current AI applications in the domain of energy supply both support and challenge decarbonization efforts. On the one hand, leveraging AI to select optimal locations for renewable energy generation or to enhance critical mineral exploration and extraction can accelerate overall energy system decarbonization efforts. On the other hand, AI applications to increase efficiency and decrease costs of fossil fuel extraction activities are in opposition to decarbonization efforts (see Figure 2 for a case study). The overall potential of AI to decarbonize energy supply is therefore not a given and rather requires explicit societal intent to meet climate mitigation targets and apply AI technologies deliberately to achieve this goal.

Critical mineral exploration is one such area where AI is already making an impact. The demand for critical minerals and metals, such as lithium, copper, and rare earth elements, is expected to triple by 2030 to meet the growing need for clean and renewable energy.³ Discovered deposits are already depleting and exploration for new sites is now expanding. However, the traditional way of critical mineral exploration takes time, and the large areas and geological processes that need to be examined add complexity and uncertainty to the process. Additionally, success rates of new site identification have been declining. AI has the potential to remove these barriers and help identify new sites for mineral exploration with more success by combining deep learning algorithms with large datasets of scientific knowledge on critical mineralization processes. For example, DeepIQ’s AI platform outperformed the most advanced industry tools (97.6% true positive rate and 20% false positive rate versus a 90% true positive rate and 25% false positive rate) when identifying new locations for critical mineral extraction in high-uncertainty contexts.³

However, at the same time, AI is also well integrated into the oil and gas industries. Major oil and gas producers are leveraging AI to enhance drilling operations, estimate reserves, and increase accuracy in exploration, hence driving down development costs. AI’s power is also being utilized to advance exploring unconventional resources, not commonly accessed in the past such as oil and gas reserves in permeable reservoirs like shale. These AI applications pose a risk to gains being made in clean and renewable energy (see section on challenges for more information).⁵

Notwithstanding these concerns about AI being used to enhance oil and gas extraction and consumption, there are also ways that AI can be leveraged to decrease the existing climate footprint of oil and gas supply. A notable example is its use to reduce or prevent fugitive methane gas leaks in oil and gas operations. By continuously analyzing data from remote sensors and satellite imagery, AI can predict or rapidly identify gas leaks. These fugitive emissions represent up to 20% of all methane emissions in oil and gas operations, therefore an application with great potential.³ Applications and services such as these are also commercially offered, for example through organizations such as GHGSTAT.

AI and Energy Transformation

AI applications in energy transformation (or generation) are the most abundantly studied applications of AI within the energy sector.⁶

AI applications in the renewable energy sector are a primary example. While renewable energy production and use have been on the rise, challenges persist. In particular, weather fluctuations and unpredictability adversely affect the reliability of renewable energy. AI can play a significant role in forecasting the intermittency of these resources, enabling more effective use of clean energy and better grid integration.^{3, 4} As buildings and neighborhoods also increasingly deploy local renewables and participate in community solar or microgrids, renewable generation forecasting at small scales is another critical application.

Conventional methods for forecasting solar and wind output primarily depend on physics-based weather prediction models, which combine observed weather conditions with process-based representations of atmospheric circulation. In contrast, AI-based approaches can incorporate a wider array of data sources and uncover subtle patterns that traditional physics-driven models often overlook.⁷ Deep learning models combining computer vision and meteorological data can significantly improve solar irradiance predictions. For example, by analyzing real-time sky images and satellite data, an AI model can predict cloud movements and solar output minutes to hours ahead with much higher spatial resolution. This allows a neighborhood solar microgrid to anticipate when clouds will reduce output, so it can pre-charge batteries or adjust loads accordingly. An important frontier in renewable forecasting is moving toward probabilistic approaches rather than relying on deterministic ones. A forecasting approach that includes confidence intervals or probabilistic scenarios is much more useful for risk-aware energy management. Deep learning architectures provide a flexible framework to generate such probabilistic forecasts by learning the distribution of the system’s output conditioned on sensor data. The result is a more robust forecasting ability, where the system not only anticipates the most likely outcome but is also prepared for less likely ones.⁷

Other emerging trends in renewable energy forecasting methods include physics-informed AI and transferable forecasting models. Physics-informed AI incorporates physical knowledge into machine learning models to improve extrapolation and reduce the data requirement for training. Transferable forecasting models refer to when a global model is trained on data from hundreds of photovoltaic (PV) systems around the world to serve as a foundation, and then fine-tuned to a specific new site (akin to transfer learning).^8, 9 AI-driven renewable forecasting is making on-site clean energy more predictable and easier to integrate. By foreseeing the fluctuations of solar and wind, building managers and automated systems can take timely actions to maintain balance, whether that is scheduling flexible loads or charging/ discharging storage. The improved foresight helps minimize the wasting of available renewable energy and reduces dependence on fossil-based backup systems, thereby making a direct contribution to decarbonization.

AI also has the potential to increase efficiency in highly complex and integrated energy systems to reduce emissions through optimal scheduling of processes. Identifying optimal strategies for coal power plants through AI has demonstrated operational efficiencies that can lead to reductions in CO₂ emissions on modelled coal power plants.¹⁰ For example, in a coal power plant model, equipped with a post combustion carbon capture, using AI to optimize scheduling led to more carbon capture and a 51% reduction in renewable energy curtailment.⁵ Carbon capture and storage in fossil fuel power plants is another energy application where AI is being used to explore more viable options to reduce emissions associated with electricity generation.³ AI can also be leveraged for predictive maintenance in power plants, where failures in energy infrastructure are forecasted and maintenance is scheduled prior to a potential breakdown.⁴

AI and Energy Consumption

AI applications to address energy consumption, and electricity consumption in particular, have great potential. In 2024, the IEA reported that electricity consumption grew at twice the average rate seen over the last decade.¹¹ Climate change impacts such as extreme weather events (heat waves, storms) are driving demand for electricity because of the increasing need for heating or cooling in buildings.¹ Additionally, industry growth, electrification of the transport sector, and digitalization are among the drivers of the growth in electricity demand.³ Building electricity use accounted for 60% of the global electricity growth in 2024.¹¹

AI can increase efficiency in energy use across buildings by enhancing methods in data clustering and energy use predictions. Modern buildings produce high-resolution streams of data through smart meters, Internet of Things (IoT) sensors, and other digital infrastructures, offering a detailed view into how and when energy is consumed. This data provides the opportunity for deep insight, particularly through AI-driven load pattern clustering, which aims to uncover structure in energy usage without relying on predefined categories. By applying unsupervised learning techniques to large-scale datasets, researchers can group days, buildings, or users based on similar consumption behaviors, revealing common load shapes, seasonal trends, and anomalies that would otherwise remain hidden.

For example, researchers have been able to uncover clear groupings among residential energy users based on their consumption behavior by applying unsupervised data-mining techniques to city-scale smart meter datasets. By learning the specific drivers behind each cluster – whether related to human behavior, environmental conditions, or system operations – AI models can predict future demand with greater precision.¹² In parallel, researchers are enriching clustering methods by integrating contextual information, such as building size, construction year, heating systems, or occupancy patterns, alongside consumption data. This multi-dimensional approach helps clarify the physical or behavioral factors behind energy use, allowing AI to link patterns with specific building characteristics. These advances contribute to a more detailed and actionable understanding of how buildings consume energy, laying the groundwork for targeted decarbonization strategies.

Building on these clustering foundations, AI is now playing an increasingly central role in predicting future energy demand. Short-term load prediction, ranging from minutes to days ahead, is crucial for operational efficiency and flexibility. Machine learning models have shown strong results in this area, outperforming classical statistical techniques by effectively modeling nonlinear influences such as weather and occupant behavior. AI has been shown to improve short-term load forecasting accuracy by over 30% when compared to generic models.¹³ Advanced AI methods, including gradient boosting and deep neural networks, can reduce forecasting errors by 20-30% in commercial buildings.¹⁴ At the neighborhood level, AI enables hierarchical forecasting, capturing demand at both individual and aggregate scales, with techniques like graph neural networks modeling relationships among buildings. This is especially beneficial in mixed-use areas, where residential and commercial loads follow different trends. Transfer learning further enhances these models’ utility by allowing knowledge from one dataset to be adapted to new buildings, even with limited historical data.¹⁵

A cornerstone of decarbonizing energy systems is harnessing demand-side flexibility, meaning the ability to shift when and how energy is consumed. An effective application of demand-side energy management is load shifting, which involves moving energy use from peak periods (when electricity is expensive or carbon-intensive) to off-peak times.¹⁶ Demand response programs aim to reduce electricity use during critical periods by offering incentives or directly controlling devices, such as remotely lowering thermostat temperatures or temporarily pausing electric water heaters.¹⁵ With advances in AI, more dynamic and optimized approaches can be applied to this task to enhance both participation and efficacy of the demand response programs.

Researchers have used reinforcement learning (RL), which refer to self-learning algorithms that optimize control strategies through continuous interaction and learning from the environment, for building load control. In an RL framework, an agent (e.g. representing a building energy management system) learns how to take optimal actions, such as adjusting thermostat setpoints, dimming lighting, or charging electric vehicles, to minimize a specified cost function (e.g., energy cost, peak electricity demand, or carbon emissions) subject to constraints (e.g., maintaining comfort).¹⁷ Recent research has demonstrated RL controllers significantly outperform fixed rule-based strategies for demand response. For example, developed an RL-based control framework for small commercial buildings that simultaneously optimizes energy cost, peak load shaving, and occupant comfort.¹⁸ By applying deep reinforcement learning to real-world building data, the system was able to independently learn how to reduce morning demand spikes and adjust operations in response to time-of-use pricing. Their model outperformed conventional heuristic control by approximately 15-25% in reducing peak demand and total energy cost, while keeping the indoor temperature within acceptable comfort ranges. This showcases AI’s ability to learn complex control strategies while adapting to changing conditions. These strategies include pre-cooling a space before a peak price interval and intelligently balancing lighting dimming with HVAC adjustments to minimize discomfort. Companies such as Uplight are already deploying AI-driven demand response management systems that aggregate thousands of homes and businesses for grid services.¹⁹ Field deployments have further shown how these AI-managed networks can respond to grid events efficiently, delivering flexibility and stability at a speed and scale that manual methods cannot match.²⁰

While the applications discussed above around AI for decarbonization show much potential, without widespread scaling and adoption, the anticipated benefits at the regional or global level will not be realized. Section 3 discusses the challenges and implications around this integration and potential scaling of AI in the energy sector.

Challenges and Enabling Conditions for AI Assisted Decarbonization in the Energy Sector

While promising real-world implementations of AI solutions exist in pilot projects and select applications, broader deployment across the energy sector remains limited – underscoring the need for expanded field demonstrations to validate performance, build confidence in the technology, and create standardized frameworks for implementation at scale.

The scarcity of high-quality, granular data required for training robust AI models across diverse operational contexts complicates large scale adoption and reliability. Another critical barrier remains: the difficulties in generalizing current approaches across varying climate zones, building types, load profiles, and shifting grid objectives, with models often trained or tested only in their specific environments. Building trust in digital technologies such as AI is another challenge. AI solutions should be explainable and transparent so that energy producers and consumers can build trust in and adopt such solutions.

A deep and persistent digital divide exists between regions of the world with inequality of access, literacy, skills, and data for digital technologies.²¹ Unfortunately, the digital divide negatively impacts those who are the most vulnerable and marginalized both in the Global North and Global South, with new forms of the digital divide taking shape (e.g., data inequity, algorithmic awareness, and quality of digital access) due to rapid technological evolution.^22, 23 Next, there are challenges in the responsible acquisition, accessibility, design, and use of appropriate AI training data and algorithms.²⁴ Furthermore, a trust deficit towards AI and other technology that has both predictive and analytical capabilities has already taken root in society and government, with some calls for outright bans on technologies such as GenAI and predictive analytics in the public sector, especially in the energy sector. More specifically, in Canadian remote communities as well as First Nations and Indigenous populations, additional dangers of AI within their energy systems emerge from data sovereignty issues, privacy concerns, and lack of appropriate AI skills training that are rooted in the contextual nuances and cultural understanding.²⁵ As a result, a casual approach to the introduction of AI within energy systems in marginalized communities can lead to significant societal and economic issues, further exacerbating already strained relationships between those communities and the federal, provincial, and local government.

While AI can help introduce and accelerate important innovations towards decarbonization, if used without caution, it can also challenge our efforts towards decarbonization and create further barriers. In addition to AI’s own growing energy consumption, the use of AI to optimize operations in heavily polluting industries, such as fossil fuels, can delay the much needed fossil fuel phaseout necessary to meet our climate goals in time.²⁶ Another potential rebound is in the building sector, where an overemphasis on demand-side efficiency improvements could lock in oil- and gas-powered building energy systems and delay the transition towards renewable systems that are ultimately needed to support decarbonization efforts. Given the rapid expansion of AI and other digital technologies, these, and other rebound effects have not yet been fully measured and mapped.²⁷

Complex problems with deploying AI in the energy sector therefore require innovative solutions and out-of-the-box thinking. Fortunately, we find some examples of innovative platforms and solutions that are responsible, open, and democratized AI for the energy and other sectors. In South Africa, companies have deployed the AI-powered Virtual Power Plant (VPP) models. VPPs create an integrated and interconnected system that can share resources (solar panels, wind turbines, storage systems, etc.) over large geographic areas with AI algorithms helping optimize and predict real-time demand and supply dynamics.²⁸ Similarly, in India AI-driven micro-grids in villages are transforming the energy delivery and consumption relationship between the government and consumers as well as establishing a self-sustaining ecosystem where communities share resources and commit to conscious usage of electricity.²⁹ Community-driven innovation to achieving the United Nations Sustainable Development Goals has gained significant momentum since the global pandemic and many communities across the world, especially those that are remote, rural, marginalized, or underserved, seek to develop bottom-up, grassroots-driven mechanisms to co-create value within their own ecosystems.^{30, 31}

The indiscriminate use of AI, however, can be dangerous and harm the very communities that AI was intended to help. For example, Canada’s First Nations Technology Council has expressed strong concerns regarding the adoption of AI systems built on Western frameworks that do not necessarily reflect or serve Indigenous community needs and are built to exclude Indigenous knowledge, values, and governance models.³² Therefore, a responsible approach to leveraging AI for vulnerable communities is essential. This responsible approach begins with inclusion of the communities in the decision-making regarding AI and engaging them in advocacy, education, and awareness about the benefits and potential dangers of AI.^{33, 34} Ensuring training data protection and addressing security and privacy concerns is another important consideration. Finally, algorithms must be developed with a “do no harm” approach where bias and prejudice are mitigated and equitable participation is offered to communities where they feel empowered rather than exploited.^{34, 35}

Despite advancements in AI, human intervention and expertise will still be necessary to support unforeseen circumstances and address ethical and other social concerns in relation to digitalization.³ Aligning AI and digital technologies with the principles of sustainable development, mitigating the negative impacts of AI, and leveraging AI to accelerate sustainable projects will help us be more deliberate with our AI adoption³⁶ (see Figure 3).

Conclusion

AI is enabling a multitude of integrated solutions to accelerate decarbonization in the energy sector, but challenges persist. In addition to the social, financial, infrastructural, and technological challenges discussed above, attention should be also given to the existing rules, power structures, and mindsets that drive our society and maintain us on a carbon intensive pathway.^{11, 37} If we do not change these rules, power structures, and mindsets, we will not be able to achieve the systematic transformations we seek, despite any progress we make in AI and other technologies.³⁷ Explicit and purposeful decisions, where solutions are implemented with a clear theory of change, are necessary by all key actors including member states, private sector, and civil society, to transition towards a zero-carbon pathway that is beneficial to all people and our planet.

About the authors

Damon Matthews is the Interim Director of Future Earth Canada and a Professor in the Department of Geography, Planning and Environment at Concordia University.

Nilushi Kumarasinghe is a consultant at Future Earth Canada and previously worked as a Research Associate at Concordia University (2022-2025).

Ursula Eicker is the Canada Excellence Research Chair in Smart, Sustainable and Resilient Cities and Communities at Concordia University and is a Professor in the Department of Building, Civil and Environmental Engineering at Concordia University.

Shahin Masoumi-Verki worked as a Research Assistant at Concordia University.

Kathryn Kaspar is a PhD Candidate at the Department of Building, Civil and Environmental Engineering at Concordia University.

Suchit Ahuja is an Associate Professor at the John Molson School of Business and the Director of the MSc Program in Business Analytics & Technology Management at Concordia University.

This article was led by Future Earth Canada, an organization working to mobilize research for action at the intersection of sustainability and digitalization. Future Earth Canada is hosted at Concordia University’s Sustainability in the Digital Age Initiative.

References

1. Clarke, L., Wei, Y.-M., De La Vega Navarro, A., Garg, A., Hahmann, A. N., Khennas, S., Azevedo, I. M. L., Löschel, A., Singh, A. K., Steg, L., Strbac, G., & Wada, K. (2023). Energy systems. In IPCC, Climate change 2022: Mitigation of climate change. Contribution of Working Group III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (pp. 613–746). Cambridge University Press. https://doi.org/10.1017/9781009157926.008
2. International Energy Agency. (2024a). Global methane tracker 2024. https://www.iea.org/reports/global-methane-tracker-2024
3. International Energy Agency. (2025). Energy and AI. https://www.iea.org/reports/energy-and-ai
4. Ferdaus, M. M., Dam, T., Anavatti, S., & Das, S. (2024). Digital technologies for a net-zero energy future: A comprehensive review. Renewable and Sustainable Energy Reviews, 202, 114681. https://doi.org/10.1016/j.rser.2024.114681
5. Chen, F., Sun, L., Jiang, B., Huo, X., Pan, X., Feng, C., & Zhang, Z. (2025). A review of AI applications in unconventional oil and gas exploration and development. Energies, 18(2), 391. https://doi.org/10.3390/en18020391
6. Alatalo, J., Heilimo, E., Rantonen, M., Väänänen, O., & Sipola, T. (2025). Reducing emissions using artificial intelligence in the energy sector: A scoping review. Applied Sciences, 15(2), 999. https://doi.org/10.3390/app15020999
7. Paletta, Q., Terrén-Serrano, G., Nie, Y., Li, B., Bieker, J., Zhang, W., Dubus, L., Dev, S., & Feng, C. (2023). Advances in solar forecasting: Computer vision with deep learning. Advances in Applied Energy, 11, 100150. https://doi.org/10.1016/j.adapen.2023.100150
8. Gijón, A., Pujana-Goitia, A., Perea, E., Molina-Solana, M., & Gómez-Romero, J. (2023). Prediction of wind turbines power with physics-informed neural networks and evidential uncertainty quantification. arXiv. https://doi.org/10.48550/arXiv.2307.14675
9. Osorio, J. D., Florio, M. D., Hovsapian, R., Chryssostomidis, C., & Karniadakis, G. E. (2025). Physics-informed machine learning for solar-thermal power systems. Energy Conversion and Management, 327, 119542. https://doi.org/10.1016/j.enconman.2025.119542
10. Ashraf, W. M., & Dua, V. (2023). Artificial intelligence-driven smart operation of large industrial complexes supporting the net-zero goal: Coal power plants. Digital Chemical Engineering, 8, 100119. https://doi.org/10.1016/j.dche.2023.100119
11. International Energy Agency. (2024b). World energy outlook 2024. https://www.iea.org/reports/world-energy-outlook-2024
12. Chen, Y., Gong, W., Obrecht, C., & Kuznik, F. (2025). A review of machine learning techniques for building electrical energy consumption prediction. Energy and AI, 21, 100518. https://doi.org/10.1016/j.egyai.2025.100518
13. Bae, H.-J., Park, J.-S., Choi, J., & Kwon, H.-Y. (2025). Learning model combined with data clustering and dimensionality reduction for short-term electricity load forecasting. Scientific Reports, 15(1), 3575. https://doi.org/10.1038/s41598-025-86982-0
14. Wang, Z., & Hong, T. (2020). Reinforcement learning for building controls: The opportunities and challenges. Applied Energy, 269, 115036. https://doi.org/10.1016/j.apenergy.2020.115036
15. Pan, S. J., & Yang, Q. (2010). A survey on transfer learning. IEEE Transactions on Knowledge and Data Engineering, 22(10), 1345–1359. https://doi.org/10.1109/TKDE.2009.191
16. Palensky, P., & Dietrich, D. (2011). Demand side management: Demand response, intelligent energy systems, and smart loads. IEEE Transactions on Industrial Informatics, 7(3), 381–388. https://doi.org/10.1109/TII.2011.2158841
17. Sutton, R. S., & Barto, A. G. (2018). Reinforcement learning: An introduction. MIT Press.
18. Madani, S., Pineau, P.-O., Charlin, L., & Desage, Y. (2025). Towards sustainable energy use: Reinforcement learning for demand response in commercial buildings. Energy and Buildings, 342, 115721. https://doi.org/10.1016/j.enbuild.2025.115721
19. Uplight. (2024). Uplight—Clean energy solutions for a decarbonized grid. https://uplight.com/
20. Johnson, L., Randhawa, S., Tyagi, P., & Walia, V. (2022, November 16). Unlocking sustainable power using Stem’s AI-driven clean energy platform on AWS. https://aws.amazon.com/blogs/industries/unlocking-sustainable-power-using-stems-ai-driven-clean-energy-platform-on-aws/
21. Dang, T. L. P., Sadreddin, A., & Ahuja, S. (2024). Readily available technologies in low-resource communities: A review and synthesis. Information Technology for Development, 30(1), 132–172. https://doi.org/10.1080/02681102.2023.2288843
22. Carter, L., Cantrell, C., & Dapeng, L. (2020). Exploring the intersection of the digital divide and artificial intelligence: A hermeneutic literature review. AIS Transactions on Human-Computer Interaction, 12(4), 253–275. https://doi.org/10.17705/1thci.00138
23. Lythreatis, S., Singh, S. K., & El-Kassar, A.-N. (2022). The digital divide: A review and future research agenda. Technological Forecasting and Social Change, 175, 121359. https://doi.org/10.1016/j.techfore.2021.121359
24. Mikalef, P., Conboy, K., Lundström, J. E., & Popovič, A. (2022). Thinking responsibly about responsible AI and the “dark side” of AI. European Journal of Information Systems, 31(3), 257–268. https://doi.org/10.1080/0960085X.2022.2026621
25. Ahuja, S., Chan, Y. E., & Krishnamurthy, R. (2023). Responsible innovation with digital platforms: Cases in India and Canada. Information Systems Journal, 33(1), 76–129. https://doi.org/10.1111/isj.12378
26. Future Earth, The Earth League, & World Climate Research Programme. (2023). 10 new insights in climate science 2023/2024. https://doi.org/10.5281/zenodo.10251586
27. Kaack, L. H., Donti, P. L., Strubell, E., Kamiya, G., Creutzig, F., & Rolnick, D. (2022). Aligning artificial intelligence with climate change mitigation. Nature Climate Change, 12(6), 518–527. https://doi.org/10.1038/s41558-022-01377-7
28. Ene, K. (2024). What role does AI play in improving Africa’s energy access? Oxford Policy Management. https://www.opml.co.uk/insights/what-role-does-ai-play-improving-africas-energy-access
29. Transitions Research. (2025). Can AI light up India’s energy shift? A pragmatic look at India’s AI and energy transitions. https://transitionsresearch.org/article/can-ai-light-up-indias-energy-shift-a-pragmatic-look-at-indias-ai-and-energy-transitions