AI for sustainability: Three steps for success

While AI holds immense promise, however, leveraging it effectively in sustainability can be complex. Moreover, the buzz around AI and sustainability often obscures their true meaning, making it challenging for organisations to grasp their actual impacts, and benefit from the synergies between them.

Integrating AI into the way you deliver your sustainability strategy can help your organisation go further, faster. To do this, here are three steps for success:

Clearly define the desired sustainability outcomes

Any organisation wanting to unlock the benefits AI can bring to sustainable organisational performance needs to start by designing an effective ‘fit for future’ sustainability strategy with clearly defined outcomes. This is not just the right thing to do, it’s the smart thing to do, as mandatory sustainability disclosures such as Streamlined Energy and Carbon Reporting (SECR), Task Force on Climate-related Financial Disclosures (TCFD), and International Sustainability Standards Board (ISSB) are increasingly prevalent, organisations will need to remain compliant with regulation.

While this article focusses on environmental sustainability (often the leading component of Environmental and Social Governance (ESG)), organisations could consider social and governance outcomes such as achieving an equal gender ratio on the company board. Specificity is key when defining these outcomes as it will strongly influence the type of AI chosen and how effective it is. For example, environmental sustainability may utilise AI in manufacturing, whilst social sustainability may use AI to remove unconscious bias in recruitment processes.

Choose the right AI solution and integrate it within the business

There are a number of areas where AI can accelerate organisations’ progress to net zero:

Use AI to increase operational efficiencies

Using AI in manufacturing or logistics can unlock multiple business benefits – by improving process efficiencies to reduce costs, boost profits, and reduce carbon emissions. For example, we recently helped a food and beverage client use AI to optimise hot drink production. By leveraging an online learning algorithm to analyse live data on beverage demand, the client was able to minimise physical waste from hot drink production and the associated energy consumption. This not only boosted profitability, but also reduced Scope 1 and 2 emissions, demonstrating the success of AI in driving multiple business benefits including environmental sustainability. This approach can be adopted by the majority of organisations in manufacturing or logistics sectors.

Use AI to improve the quality and quantity of emissions data

To assess their progress towards net zero, organisations need up-to-date information on their carbon outputs and offsets. However, the methods of calculating emissions vary (for example, energy consumption from buildings uses different calculation inputs from employees' carbon footprint) meaning emissions data is often difficult or time consuming to capture. This is where AI can help. Organisations such as Treeconomy use AI models to process high-resolution remote sensing data that quantifies and classifies the different habitat types at a given area. This allows companies to measure the amount of carbon their open spaces remove, providing accurate data to support their net zero strategies. Organisations without open habitats could use AI to automatically collect live data from their production line (for example, the energy consumption of a machine) to make informed decisions on which parts of their supply chain are responsible for their carbon emissions.

Use AI to improve regulatory compliance

Large organisations are often required to report on multiple sustainability policies (such as SECR and TCFD) which take up valuable time and resources. Generative AI could quickly produce reporting frameworks by drawing on data and text from across the business, before being reviewed by humans. This would enhance the benefits of accurate emissions data – as policy makers can receive data in a way that doesn’t burden organisations.

While sustainability and AI are sometimes framed as competing directions for a company to pursue, they are often mutually beneficial. Deploying AI in an area of the organisation where it can deliver multiple benefits, as highlighted earlier by the optimisation from our food and beverage client, means it can achieve this strategic alignment.

Evaluate the sustainability impact of AI technologies

There are different types of AI technologies that organisations can implement, ranging from object detection to image classification. Whilst they may seem similar, there are notable variances in the emissions produced from each type. For instance, image generation can produce up to 200× more CO2e than text classification. Organisations should collate and scrutinise robust data to make informed decisions about the true emission impacts of AI. Typically, this involves calculating the emission savings resulting from AI efficiency gains (for example, reduced electricity or travel) and subtracting the emissions produced from the AI (the power used to provide the AI multiplied by the emission conversion factor). It is important to consider the environmental footprint of AI implementations as implementing the wrong type could negate the carbon reductions you aim to achieve.

AI’s potential to revolutionise markets and business models has made it a hot topic. Whilst it can offer serious sustainability benefits to organisations and operations, it can also be expensive, complex, and counterproductive if not used in the right context. To get the best out of AI to accelerate sustainability goals, organisations must think of AI and sustainability as complementary rather than competing growth drivers.