Sust-AI-nability: Sounding out the future of Artificial Intelligence and its role in sustainability

'Artificial Intelligence’. If there was a buzzword to learn in 2023, this was it. But what is AI really, and what does it mean for the future of companies’ sustainability journeys?

I spoke with Justin Tan, Founder and Managing Director of Evolutio Consulting, an innovative advisory and training company focused on helping individuals, businesses and organisations to understand and implement AI and other disruptive technologies, to hear his thoughts on how AI will shake up the sustainability space and, more importantly, why business leaders should care.

Can you share with us your journey in the field of AI and how it led you to establish Evolutio Consulting?

My journey into AI is deeply rooted in my experience of over 15 years in the digital, data and innovation sectors, starting at Kearney, where I guided companies through their digital and data transformation journeys. Subsequently, as Director of Innovation at MSD (also known as Merck & Co.), I collaborated with start-ups to introduce cutting-edge technologies to patients, including a cornerstone project involving the deployment of an AI-enabled oncology patient monitoring platform, which further developed my in-depth engagement with AI.

These experiences led me to found Evolutio Consulting, with a vision to bridge the gap between AI's potential and its practical application in business settings. Beyond providing strategic advisory and implementation services, we also emphasise the importance of executive education and training. This commitment to fostering AI literacy and proficiency reflects my broader mission to help businesses not only adapt to the technological landscape but to thrive within it, and actively shape it for the betterment of organisations and the communities they serve.

How do you approach the dichotomy of AI's environmental footprint versus its role in promoting more sustainable business practices?

On the one hand, AI’s environmental footprint cannot be overlooked. Direct impacts include water and electricity consumption for training and running models and building and operating data centres. Indirect effects encompass resource extraction required for manufacturing GPUs (the main chips used for training AI models) and other AI infrastructure, and costs arising from increased data transmission, storage, redundancy, and replication, all of which are by-products of the boom in AI-generated content.

These impacts are exacerbated by the dominant approach to developing AI over the last 20 years. This has focused on boosting output by increasing the size of training data sets and building bigger ‘AI brains’ rather than improving the ability of existing ’brains’ to learn more efficiently. The result is models of increasing size and energy-intensity. While the latest AI models reflect efficiency innovations in training methods, model architectures, and algorithms, the trend towards larger (instead of more efficient) models is likely to continue for several more years.

On the other hand, AI equally holds massive potential to support sustainability by enhancing infrastructures and operations in terms of efficiency (e.g. predictive maintenance systems, smart grids, intelligent traffic management systems), precision (e.g. precision agriculture which minimises water and fertiliser consumption), and monitoring (e.g. deforestation and land use change detection). Its potential applications in this regard are vast and still massively untapped. PwC for instance, estimates that AI could help reduce greenhouse gas emissions by 1.5-4.0% by 2030. My view is that improving energy efficiency represents one of the lowest hanging fruits, with some studies estimating AI-driven efficiencies of 10-40%.

Is the promise of AI’s role in promoting more sustainable business practices realistic?

While discussions of AI's impact on sustainability often take a black or white perspective, it is, in fact, a highly complex and multifaceted issue. The devil is in the detail, and it's crucial to recognise that potential alone does not guarantee achievement.

One of the nuanced considerations is AI’s ‘black box’ nature. Even AI developers often do not fully understand how models arrive at their outputs. This poses a significant challenge for applying AI in environmental contexts, where decisions can have far-reaching consequences on ecosystems, climate, and human communities, and where trust and accountability are therefore paramount. Mechanisms for transparency and interpretability will be essential to ensure that AI's application in environmental contexts upholds ethical standards and is accountable to all stakeholders.

The role of regulation and legislation is clearly becoming pivotal. Since AI development and usage result in negative externalities, the notion of taxing AI, potentially according to energy intensity, should be on the cards. Similarly, efforts to galvanise governmental support and funding for environmentally beneficial AI technologies and applications are needed, especially in relation to public goods such as wildlife conservation, where commercial incentives may be insufficient to drive investment and R&D. Currently however, most AI policy and regulatory discussions focus on balancing public safety while boosting innovation, often overlooking environmental implications.

The discussion of AI's environmental impact is further complicated by issues of inequality. The concentration of AI development and benefits in wealthier nations risks exacerbating global disparities. Many developing countries, with rising environmental and greenhouse gas emissions, continue to lack the means or incentives to switch from coal- to oil-based power generation. Is the introduction of AI technologies even relevant for such economies over a 5 to 10-year horizon? It would likely require the US, Europe, and China, as leaders in the field, to commit to a combination of technology transfer, funding mechanisms, and other forms of global cooperation for AI to be in play for developing countries.

Finally, we must also consider AI in the context of other competing technologies and initiatives that could also benefit the environment. With AI being the ‘new kid on the block’ (at least in relation to sustainability), most studies about AI’s impact remain high-level and theoretical. Consequently, more comprehensive, and comparative impact assessments are required to understand both potential benefits and costs relative to other existing technologies and options before making decisions on its applications.

For companies looking to start their AI journey, what initial steps do you recommend for ensuring successful implementation?

Generative AI is the natural starting point for many businesses, as this variant of AI is more accessible and with broader applications than ‘traditional’ AI. For most businesses, there are three critical prerequisites for implementing Generative AI.

Firstly, and most importantly, businesses need to invest in building up AI literacy and proficiency so their teams understand and can consider applications within their own workflows, allowing the business to identify and harness the full potential of the technology.

Secondly, companies also need to ensure their teams can safely and responsibly leverage the technology by having a clear AI usage policy and guardrails that are well communicated, with stakeholder buy-in, across the organisation.

Thirdly, businesses should adopt a ‘people-first’ approach to AI. AI comes with a lot of ‘baggage’ such as ethical considerations around privacy and bias, and fears the technology is on track to replace jobs. The role of leaders and management in being empathetic, and shaping direction and narrative is therefore critical, and not simply a ‘nice-to-have'.

Once these building blocks are in place, businesses can then consider experimenting with more complex AI use-cases, while incrementally building the capabilities required to move their organisation from testing and learning, to building and scaling AI applications.

------------------------------

Justin Tan is the Founder of Evolutio Consulting, an innovative advisory and training company dedicated to helping individuals, businesses, and organisations to understand and implement AI and other disruptive technologies. He believes in a people-first and capability-led approach to technology transformations, supporting both individuals and organisations by designing and delivering tailored AI training programmes that emphasise practical, hands-on learning. He has partnered with firms of all sizes and from a range of sectors, helping to prepare them for a future where technology and AI play a central role in their growth and success.