The rapid advancement of AI has been transforming industries, but it’s also casting a looming shadow over the planet’s energy crisis and it’s something that organizations cannot ignore. 

In this Q&A, Joseph Yang, Managing Director at HPE Singapore gives insights on AI and the need for organizations to integrate sustainability into every facet of the AI ecosystem.

Joseph Yang

Managing Director at HPE Singapore

  1. Can you briefly discuss how sustainability comes into the picture as companies now begin embracing AI?

Yang: AI is becoming an integral component of our future, already making disruptive changes across industries and different aspects of society. However, in the rush to apply AI, companies run the risk of increasing their carbon footprint, as the amount of energy required to use AI – from training to tuning to inference – is enormous and will continue to grow.

As companies are under increased scrutiny and pressure from external stakeholders to fast-track net-zero targets and improve sustainability impacts, they must embed sustainability into their AI strategies from the start and address the energy consumption of their AI models at every level.

A crucial part of AI sustainability is optimizing the use of resources – the faster the supercomputer, the more efficient use of time and energy is deployed to run AI models.

2. What are the significant energy consumption of AI and carbon emissions associated with AI models, resulting in their contribution to climate change?

Yang: Large AI models require massive computing power and energy to train and run. According to researchers from the University of Massachusetts, Amherst, the model training process for natural-language processing could emit five times as much carbon dioxide as the lifetime emissions of an average American car.

Tuning AI models, in particular, is the most computationally intensive stage, producing the most carbon and consuming the most power. It is, then, crucial for companies to strive towards energy efficiency in all aspects of AI model development and usage, including building energy-efficient high-computing systems and seeking opportunities to purchase renewable energy where possible. An example of this is the HPE GreenLake for Large Language Models that we announced earlier this year, which runs on supercomputers initially hosted in QScale’s Quebec location and provides power from 99.5% renewable sources.

3. Can you give specific examples (companies, for instance) of entities that are now using sustainability strategies as they continue to integrate AI into the business?

Yang: While running AI models requires significant energy consumption, AI as a technology has the potential to help tackle climate change and help businesses across a wide range of industries improve environmental impacts thanks to its pattern recognition, predictive analytics, and process optimization capabilities.

For instance, in retail, AI algorithms can help stores optimize logistics, predict demand, and minimize wastage as well as enhance energy management. AI can also transform supply chains by analyzing data from every stage of the process to spot inefficiencies, enabling businesses to make better decisions to improve energy efficiency.

For example, Carnegie Clean Energy is using HPE’s supercomputer technology and AI to help unleash the renewable energy stored in ocean waves. 

4. Related to the first question, how can sustainability be integrated into every aspect of the AI ecosystem? And what are the possible challenges that companies or organizations face when integrating sustainability and AI?

Yang: Integrating sustainability seamlessly into AI adoption for businesses entails a comprehensive approach that first begins with responsible data collection that upholds accuracy, respects privacy, and operates on sustainable energy sources. Prioritizing the creation of efficient algorithms, using resources judiciously, and minimizing waste is also crucial.

These algorithms should be designed to align with sustainability objectives, promoting eco-friendly practices and considering the broader impact on the environment and society. Additionally, AI models can also optimize supply chain management, allowing businesses to significantly reduce waste, minimize energy consumption, and in turn, lower the overall carbon emissions associated with logistical operations.

However, implementing sustainability into AI practices is not without its challenges. Businesses face issues with access to comprehensive and relevant data that can provide intelligent insights into sustainable practices, free from biases and inaccuracies. Furthermore, businesses also face challenges in hiring and upskilling professionals who are equipped with the expertise and knowledge in AI technologies, models, and algorithms.

5. How can C-suite leaders integrate sustainability goals into their overall business strategy?

Yang: C-suite leaders can integrate sustainability goals into their overall business strategy through a strategic and holistic approach that encompasses various aspects of the organization. Firstly, ensure that sustainability goals align with your organization’s vision, values, and long-term objectives. Sustainability should become an integral part of the company’s identity and purpose. It’s important to involve stakeholders, including employees, customers, suppliers, and communities, in setting sustainability goals, as well as collaborate with external partners and other organizations to enhance collective impact and knowledge sharing.

Additionally, adequate allocation of resources and investments, establishment of relevant KPIs, and integration of sustainability into day-to-day operations are crucial as well.