Greening AI and 6G-V2X: Pioneering Sustainable Innovation in Artificial Intelligence

The rapid proliferation of artificial intelligence (AI) technologies has ushered in an era of unprecedented innovation, transforming industries from healthcare to transportation. However, this technological leap comes with significant environmental costs. The computational demands of AI, particularly in training large-scale models, contribute to substantial energy consumption and carbon emissions. As the global information technology (IT) sector is projected to consume up to 21% of the world’s electricity by 2030, the urgency to address AI’s ecological footprint has never been greater. In response, academic initiatives worldwide are tackling these challenges head-on, aiming to harmonise AI’s transformative potential with environmental and infrastructural sustainability.

Two prominent efforts—the “Greening AI with Software Engineering” workshop held in Lausanne, Switzerland, from 3–7 February 2025, and ongoing global research into AI-integrated 6G vehicle-to-everything (V2X) communications—exemplify this commitment. These initiatives underscore a pivotal shift towards sustainable AI, leveraging software engineering principles and network optimisation to mitigate environmental impacts while maintaining performance. This article explores these groundbreaking academic efforts, delving into their objectives, methodologies, and implications for the future of AI. By examining the intersection of software engineering, network optimisation, and sustainability, we highlight how researchers are shaping a responsible AI ecosystem that balances innovation with ecological stewardship.

The “Greening AI with Software Engineering” Workshop: A Landmark Event

Overview and Objectives

The “Greening AI with Software Engineering” workshop, held in Lausanne and co-funded by the Centre Européen de Calcul Atomique et Moléculaire (CECAM) and the Lorentz Center, convened 29 interdisciplinary participants, including academics and industry practitioners, to address the environmental impact of AI-enabled systems. Held from 3–7 February 2025, the event aimed to advance green software and AI research by fostering collaboration and identifying key challenges in developing sustainable AI solutions.

The workshop’s primary objective was to integrate software engineering principles into AI development to enhance energy efficiency and reduce carbon footprints. Participants explored strategies such as optimising AI model training, designing lightweight algorithms, and minimising computational waste. By prioritising sustainability without sacrificing performance, the initiative sought to create a roadmap for environmentally responsible AI practices.

Key Focus Areas

The workshop’s agenda was structured around six critical focus areas, each addressing a facet of sustainable AI development:

1. Energy Assessment and Standardisation

Participants emphasised the need for standardised metrics to measure AI systems’ energy consumption. Current methodologies vary widely, hindering comparability and reproducibility. Proposals included community-driven oversight mechanisms, such as a metrics committee, to curate and endorse measurement practices.

2. Benchmarking Practices

Establishing benchmarks for energy-efficient AI models was a priority. Unlike traditional benchmarks focused on accuracy, these incorporate energy consumption and carbon emissions as key performance indicators (KPIs).

3. Sustainability-Aware Architectures

Researchers explored architectural designs that inherently reduce energy demands, such as modular or adaptive AI systems that scale resources dynamically based on task complexity.

4. Runtime Adaptation

Techniques like dynamic model pruning and adaptive inference were discussed to enable AI systems to adjust computational requirements in real time, minimising energy use during operation.

5. Empirical Methodologies

Drawing from empirical software engineering, participants advocated for robust, reproducible studies to validate sustainability claims. This includes case studies and surveys to assess real-world impacts.

6. Education and Awareness

The workshop highlighted the need to integrate green AI principles into computer science curricula, equipping the next generation of developers with skills in sustainable software engineering. Plans were proposed to create a global hub for sharing educational resources.

Workshop Format and Outcomes

The workshop’s format was designed to encourage co-creation and active participation. It featured one or two keynote speeches daily, followed by flash talks, panel discussions, hands-on sessions, and collaborative working groups. This structure facilitated meaningful dialogue, allowing participants to refine focus areas through iterative processes like sticky-note brainstorming and rotating group discussions.

Key outcomes included a comprehensive research agenda outlined in the workshop’s report, published on arXiv. The report synthesises insights from participants, proposing actionable strategies for industry and academia. It also catalysed the formation of an interdisciplinary community dedicated to advancing green AI, with plans for ongoing collaboration with industry partners and environmental advocacy groups.

Significance and Global Context

The Lausanne workshop aligns with broader global efforts to address AI’s environmental impact. For instance, the Green Software Foundation’s initiatives, such as the Software Carbon Intensity (SCI) specification, promote transparency in software emissions. Similarly, Singapore’s Green Computing Funding Initiative (GCFI) supports research into energy-efficient software, serving as a model for public-private partnerships.

By integrating software engineering with sustainability, the workshop addresses a critical gap in AI research. Traditional AI development prioritises accuracy and performance, often at the expense of energy efficiency. The Lausanne initiative challenges this paradigm, advocating for a holistic approach that considers the entire software lifecycle—from design to deployment and maintenance.

6G-V2X AI Surveys: Pioneering Sustainable Wireless Networks

The Evolution to 6G and V2X Communications

As AI applications expand into real-time domains like autonomous driving, the demand for robust, low-latency wireless networks has surged. The transition from 5G to 6G networks represents a strategic response to this demand, promising ultra-high-speed, ubiquitous connectivity. Vehicle-to-everything (V2X) communications, which enable vehicles to interact with other vehicles, infrastructure, pedestrians, and networks, are central to this evolution.

6G-V2X systems leverage AI to enhance functionality, supporting applications such as cooperative perception, real-time traffic management, and predictive maintenance. However, integrating AI into these systems poses significant challenges, including increased energy consumption and bandwidth demands. Global researchers are conducting surveys to address these issues, focusing on sustainable network optimisation techniques.

Research Objectives and Methodologies

The primary objective of 6G-V2X AI surveys is to develop AI-integrated networks that deliver high performance while minimising environmental and infrastructural impacts. These studies employ a range of methodologies, including:

1. Literature Reviews: Comprehensive analyses of existing research on AI in 5G and 6G networks, identifying gaps in energy-efficient practices.

Simulation-Based Studies: Modelling 6G-V2X scenarios to evaluate AI algorithms’ energy consumption and performance under varying conditions.

2. Field Experiments: Testing AI-driven V2X applications, such as edge-based autonomous driving, in real-world environments to assess scalability and efficiency.

3. Key research questions include: How can AI optimise network resource allocation without escalating energy use? What are the trade-offs between latency, accuracy, and sustainability in AI-driven V2X systems

Key Technical Approaches

Researchers are exploring several technical approaches to achieve sustainable 6G-V2X systems:

1. Edge Computing: By processing AI tasks at the network edge, edge computing reduces latency and bandwidth demands, lowering energy consumption compared to cloud-based processing

2. Smart Routing: AI algorithms optimise data transmission paths, minimising energy-intensive retransmissions and congestion. Techniques like reinforcement learning enable adaptive routing based on real-time network conditions. 

3. Adaptive Communication Protocols: These protocols dynamically adjust transmission power and frequency to match application requirements, reducing energy waste. For example, in low-traffic scenarios, protocols can lower power levels without compromising connectivity.

4. Energy-Efficient AI Models: Lightweight AI models, such as compressed neural networks, are designed for resource-constrained V2X devices, balancing performance with energy efficiency. 

5. Green Federated Learning (GFL): GFL enables distributed AI training across V2X devices, keeping data local to reduce communication overheads. By optimising model updates, GFL minimises energy use while maintaining accuracy.

Challenges and Opportunities

Despite their promise, 6G-V2X AI systems face several challenges:

1. Energy Consumption: AI models, especially large language models (LLMs), are computationally intensive, straining network resources.

2. Scalability: Ensuring AI-driven V2X systems scale to millions of devices without compromising efficiency is a complex task.

3. Interoperability: Standardising protocols across diverse V2X ecosystems is critical to seamless integration.

However, these challenges present opportunities for innovation. For instance, blockchain and digital twins—advanced technologies explored in 6G research—can enhance security and efficiency in V2X networks. Additionally, the global push for sustainability, exemplified by the United Nations’ Sustainable Development Goals (SDGs), provides a framework for aligning 6G-V2X research with environmental objectives.

Case Studies and Real-World Applications

Several projects illustrate the practical impact of 6G-V2X AI research:

1. European 6G Initiatives: The European Union’s 6G Flagship programme, Hexa-X, explores AI-driven network optimisation for V2X applications, focusing on energy efficiency and low-latency communications.

2. Chinese 6G Trials: China’s Three-Year Action Plan on New Data Centres integrates AI into V2X networks, prioritising sustainable data centre operations.

3. UAV-Based V2X : Research on unmanned aerial vehicles (UAVs) as mobile V2X nodes demonstrates how AI can optimise drone-assisted communications, reducing energy consumption in remote areas.

These initiatives highlight the transformative potential of AI in 6G-V2X systems, from enhancing autonomous driving safety to optimising urban mobility.

Synergies Between Greening AI and 6G-V2X Research

Shared Goals and Methodologies

The “Greening AI with Software Engineering” workshop and 6G-V2X AI surveys share a common goal: to develop AI systems that are powerful, intelligent, and sustainable. Both initiatives employ software engineering and optimisation techniques to address energy efficiency, drawing on empirical methodologies and interdisciplinary collaboration.

For example, the workshop’s emphasis on lightweight algorithms aligns with 6G-V2X research into compressed AI models for edge devices. Similarly, the workshop’s focus on runtime adaptation parallels 6G-V2X efforts to develop adaptive communication protocols.

Complementary Impacts

The two initiatives complement each other in their scope and impact:

1. Greening AI Workshop: Provides a broad framework for sustainable AI across domains, applicable to 6G-V2X systems. Its focus on software lifecycle integration ensures long-term sustainability.

2. 6G-V2X Surveys: Offer domain-specific insights into real-time AI applications, addressing unique challenges like latency and scalability. These findings can inform broader green AI practices.

Together, they create a synergistic approach, combining general principles with application-specific solutions to advance sustainable AI.

Potential for Collaboration

There is significant potential for collaboration between these initiatives. For instance, 6G-V2X researchers could adopt the workshop’s standardised energy metrics to benchmark network performance. Conversely, the workshop’s educational hub could incorporate 6G-V2X case studies to train developers in domain-specific sustainability practices.

Broader Implications for Sustainable AI

Environmental Impact

By reducing AI’s energy consumption and carbon emissions, these initiatives contribute to global climate goals. The IT sector’s projected growth underscores the urgency of these efforts. For example, training a single large-scale AI model like GPT-3 can emit 550 tons of CO2, equivalent to multiple transcontinental flights. Sustainable AI practices mitigate these impacts, aligning with policies like the European Code of Conduct for Data Centres.

Economic and Social Benefits

Sustainable AI offers economic benefits, such as reduced operational costs through energy-efficient systems. Socially, it promotes equitable access to technology by lowering resource demands, enabling deployment in resource-constrained regions. The Lausanne workshop’s focus on social impact, through interdisciplinary discussions with social scientists, underscores this commitment.

Policy and Industry Alignment

These academic efforts align seamlessly with broader industry and policy initiatives aimed at mitigating the environmental impact of digital technologies. Leading tech companies such as Google and Microsoft have made bold pledges to operate entirely on carbon-free energy by 2030. These commitments are not merely symbolic—they represent a shift in how technology companies view their role in climate responsibility. Google, for instance, is investing in advanced computing infrastructure that relies on clean energy sources, as well as AI tools that help optimise energy usage in its data centres. Microsoft has committed to becoming carbon negative, meaning it will remove more carbon from the atmosphere than it emits, and is actively developing sustainability metrics to hold itself accountable.

At the governmental level, efforts such as Singapore’s Green Data Centre Roadmap demonstrate how policy can drive industry transformation. This initiative outlines clear sustainability benchmarks for data centres, including targets for energy efficiency, carbon emissions, and the use of renewable resources. It encourages the development and deployment of innovative cooling technologies, AI-based energy management, and the integration of green building practices.

In this landscape, academic research plays a critical role by offering the theoretical frameworks and empirical insights needed to guide and scale these efforts. Universities and research institutions contribute cutting-edge work on algorithmic efficiency, green software engineering, and the carbon impact of AI systems. Their findings support the development of robust lifecycle assessments and inform sustainability metrics that companies and regulators can adopt. Moreover, academia acts as a bridge for public-private collaboration, facilitating knowledge transfer and joint innovation between researchers, industry leaders, and policymakers.

By aligning goals and sharing knowledge, these collaborative efforts help accelerate the transition to a more sustainable digital future—one where technological innovation thrives without compromising planetary health.

Challenges and Future Directions

Technical Challenges

1. Measurement Consistency: Developing universal metrics for AI energy consumption remains a hurdle.

2. Trade-Offs: Balancing accuracy, latency, and sustainability requires sophisticated optimisation techniques.

3. Scalability: Ensuring sustainable practices scale to large-scale deployments is critical.

Research Gaps

1. Longitudinal Studies: More research is needed to assess the long-term impact of green AI practices.

2. Interdisciplinary Integration: Bridging software engineering, network optimisation, and environmental science requires sustained collaboration.

3. Emerging Technologies: The sustainability of quantum computing and generative AI needs further exploration.

Future Directions

Future research should focus on:

1. Automated Tools: Developing AI-driven tools to optimise energy use in real time.

2. Global Standards: Establishing international standards for green AI and 6G-V2X systems.

3. Education: Expanding curricula to include sustainability as a core component of AI development. 

The “Greening AI with Software Engineering” workshop and the 6G-V2X AI surveys represent a pivotal moment in the global effort to create a more sustainable and responsible future for artificial intelligence. These forward-thinking initiatives aim to tackle one of the most pressing concerns of our time: the environmental and infrastructural toll of AI’s rapid growth. As the demand for computational power continues to surge—particularly with the advent of large-scale AI models and ever-connected smart devices—the need for energy-efficient, scalable, and sustainable systems becomes increasingly urgent.

By leveraging the principles of software engineering and network optimisation, both initiatives offer a promising roadmap for minimising AI’s carbon footprint. The workshop brings together software engineers, AI researchers, policymakers, and sustainability experts to explore new ways of designing AI systems that are not only powerful but also environmentally conscious. Topics such as energy-efficient algorithms, green software development practices, and lifecycle carbon analysis are at the heart of these discussions. The collaborative environment fosters a culture of shared responsibility and collective problem-solving, encouraging open-source solutions and industry-wide standards.

Meanwhile, the 6G-V2X AI surveys delve into the intersection of next-generation communication technologies and AI-driven vehicle-to-everything (V2X) ecosystems. As autonomous vehicles and smart transportation systems rely heavily on AI for real-time decision-making, ensuring the sustainability of such infrastructure is critical. These surveys assess how AI can be embedded into 6G networks in a way that maximises performance while minimising energy usage. Key areas of focus include edge computing, intelligent routing protocols, energy-aware AI architectures, and the integration of renewable energy sources within communication systems.

What sets these initiatives apart is their interdisciplinary approach, combining academic rigour with practical application. They serve not only as research platforms but also as catalysts for policy change and educational reform. By involving stakeholders from across academia, industry, and government, they ensure that solutions are both scalable and impactful.

As AI continues to reshape economies, industries, and societies, these efforts underscore a vital truth: technological advancement must go hand-in-hand with environmental stewardship. Only by fostering collaboration, standardising best practices, and educating the next generation of engineers and developers can we ensure that AI’s evolution remains aligned with the planet’s ecological limits. In doing so, we lay the groundwork for an AI-driven future that is not only intelligent and innovative but also sustainable, inclusive, and just