By Dorian Selz, CEO & co-founder, Squirro
As companies race to deploy generative AI, they’re facing an ironic challenge: the very technology designed to drive efficiency is jeopardising sustainability goals. The importance of ESG (Environmental, Social, and Governance) on the corporate boardroom agenda is widely acknowledged. It helps to drive sustainable growth, risk management, investor interest and stakeholder trust, and in recent years it has become an important indicator of a company’s ethical performance and brand values. Integrating ESG principles can help organisations to address climate change concerns and meet regulatory compliance, but the explosion in AI adoption has thrown a spanner in the works, evolving so quickly that ESG reporting procedures haven’t kept pace.
The irony is that GenAI usage is significantly increasing corporate energy consumption thanks to the growing power demands of training and running large language models. Both Google and Microsoft admitted last year that energy use related to the data centres supporting their AI strategies is endangering their plans to become carbon negative. It’s not just the growth of the technology that has evaded ESG reporting; there is also the fact that AI depends on third-party cloud services, making it challenging to calculate and accurately reflect all the components, including electricity, water, rare elements and data centre infrastructure, in sustainability reports.
A question mark currently hangs over how stringent regulators will be at adapting ESG reporting frameworks as GenAI usage evolves. Large corporate organisations, not just in tech, but other industries too, are scaling back their DEI (Diversity, Equity, Inclusion) initiatives with the encouragement of the new US administration. This could set precedents for other corporate commitments, leading to a delay in standardising how the impact of AI is measured and reported.
In the financial services sector, where a cautious approach stemming from complex compliance requirements and risk management priorities tends to inform all regulatory matters, there will undoubtedly be a tendency to continue developing enterprise GenAI applications with ESG principles in mind. The real question is what this will look like in practice?
Below are five routes that finservs can take to ensure GenAI enables ESG rather than undermines it. That way the company benefits, as does the broader ecosystem in which it sits.
Channelling GenAI with ESG at the core
1. Work towards AI being part of ESG strategy
Addressing the environmental impact of AI is crucial, and efforts to mitigate emissions, such as partnering with data centres that are committed to using renewables for their energy sends a clear and positive signal to investors and the public. AI should be incorporated as part of an umbrella ESG strategy, demonstrating leadership in technological innovation as well as in corporate sustainability.
2. Balancing performance with sustainability
The energy requirements of large language models (LLMs) are huge. According to a report from Goldman Sachs, one ChatGPT query demands almost 10x the electricity of a standard Google search. Finservs should reflect on how they can balance the performance and sustainability of their AI training models and GenAI deployments.
3. Select the right AI model
In many cases AI tasks can be accomplished using less complex, lower energy demanding models without compromising on performance. If a smaller language model (SLM) can achieve an objective just as efficiently as a larger model, it would be sensible to opt for this to ensure sustainability objectives (as well as costs) are met. Time spent scoping AI-enabled tasks will help to avoid unnecessary energy consumption and reduce carbon footprints.
4. Stay agnostic and avoid lock-in
AI will evolve quickly which will introduce new opportunities and a changing regulatory landscape. Financial services companies can ensure they are prepared for future-proofing AI deployments if they use LLM-agnostic solutions that do not tie them into a specific AI technology. This will allow them to pivot easily as AI strategies change and more energy-efficient models come to market, allowing them to be compliant with future ESG requirements.
5. Offset AI with other gains
There are multiple ways that companies can become more efficient using AI to offset the environmental footprint associated with their use of the technology. AI is ideal for optimising supply chains, implementing and improving automated tasks, boosting efficiency and cutting wastage and energy consumption across various operations.
Next steps to align AI and ESG
Financial services companies can utilise GenAI platforms designed specifically to align AI deployments of all sizes with ESG goals while ensuring maximum performance.
When specifying a platform, it’s important that companies look for one that provides an LLM-agnostic approach which will give them the autonomy to deploy future AI advances without being tied to a single model or vendor and allow them to run LLMs locally on their own chips in their own environment. Ideal platforms will use retrieval-augmented generation (RAG) instead of retraining large models, integrating new data from data repositories seamlessly, cutting compute demand and power requirements.
Platforms need to offer cutting-edge data chunking and vectorisation techniques which reduce how much information needs to be processed by the LLM, subsequently boosting computational efficiency, and minimising the emissions associated with AI.
When it comes to deployment, it’s important to work with a solution that can support Virtual Private Cloud, on-premises or a hybrid model. Finservs are likely to want customisable configurations and may need geographical as well as environmental requirements to be taken into consideration. If the solution also offers simplified compliance management and ESG reporting that can leverage regulatory documentation and enterprise data, this will save time and resources for the organisation.
Given the tumultuous impact of GenAI adoption and the uncertainty regarding ESG reporting, financial services companies are at a crossroads, but it is important to focus on the benefits that these technology advancements deliver. Determining how AI will be used, selecting the most appropriate AI model and planning for efficient deployment are essential and can be facilitated with dedicated GenAI platforms. Perhaps the most important role they can play, however, is in automating core business processes that drive efficiency and save resources, taking companies closer to their sustainability goals. By aligning AI strategies with ESG principles, they can drive both technological innovation and sustainable growth. The future of finance isn’t just digital—it’s responsible.”
