Generative AI and ESG: 3 steps to chart a sustainable course

By  Chris Kirkpatrick

As companies explore the ways generative AI can transform customer and employee experiences, some may overlook the question of how well this technology aligns with their environmental, social and governance (ESG) goals.  

A recent study found that creating a single image with an advanced Al model consumes as much energy as fully charging a smartphone.¹ Another study suggests Chat GPT-3 "drinks" the equivalent of a 500ml bottle of water for every 10-50 responses it produces.²

Now, let's scale that up.

Imagine the cumulative footprint when these tools are employed across an entire multinational organization. That’s an environmental impact that companies today cannot afford to overlook as they consider their long-term strategic business objectives holistically.

Current generations in the workforce aren’t shy about voting with their feet, and taking business elsewhere if a company's values don't align with theirs.³ Meanwhile, regulatory initiatives like the Corporate Sustainability Reporting Directive (CSRD) set the stage for more standardized and enforceable corporate ESG practices.⁴

So how can your team execute a successful generative Al play without losing sight of these stakes?

Shift from an 'ESG' to a 'SEE' mindset

My day-to-day focuses on elevating employee and customer experiences. This has led me to a perspective shift when it comes to the question of sustainability: from an ESG to a SEE (social, economic, environmental) mindset.

The SEE mindset prioritizes ethical and social considerations. In the context of generative Al, it means evaluating the technology’s broader impact, beginning with its effect on your employees and customers. This might seem like a subtle pivot, but I’ve found that approaching sustainable practices with a people-first mindset is a powerful way to motivate and drive change. After all, the true value of a cleaner planet lies in ensuring it can be enjoyed by future generations.

When viewing generative Al opportunities through a SEE lens, the initial questions a team would ask are:

1. What is the potential social impact? How does using generative Al affect employee and customer experience? What are the implications for the brand and reputation?
2. What is the economic impact? What are the potential financial and risk implications of scaling up generative Al tools? How might it influence our business outcomes?
3. What is the potential environmental impact? How does the value of this technology weigh against its emissions footprint? What is its effect on our environmental objectives, like achieving net zero.

How to chart a sustainable course for generative Al through SEE

Here are three steps for any team aiming to find this balance and maximize the benefits of generative Al, while aligning with company values:

Step 1: Establish and measure your baseline

From day one, make it a priority to integrate a system of measurement and monitoring, so you know exactly how the technology drives efficiency and experience among your workforce, as well as its influence on your energy usage and its broader environmental impact.

With this practice in place, the next phase is to understand and document across important baselines before implementing your generative AI solution. Without a thorough understanding of your existing, A) greenhouse gas emissions, B) quality of existing employee and customer digital experiences, and C) current business risk profile and dependency on energy and third-party services, it becomes very difficult to assess the true impact of generative Al on these areas.

Step 2: Set thresholds and targets that meet wider business objectives.

With any emerging technology, it’s all too easy to develop tunnel vision, channeling focus and resources into a single objective at the expense of equally critical goals.

To address this, key stakeholders from across your different lines of business should consider sitting down for a sync about their respective thresholds and targets. The objective here is more than just a familiarity with other departments’ goals. Rather, it’s to ensure your team understands how these various goals impact each other in both the short and long term.

Naturally, this process will involve negotiation, prioritization and compromise. Not every threshold or target can lead the charge. Some teams, for example, may decide to place a premium on automation, while remaining fully aware that this will require plans to offset emissions in the future. For others, it may mean prioritizing employee experience, while recognizing the need to cut costs again down the line.

Step 3: Create holistic business outcome measurements to keep on track

As teams roll out their first generative AI use cases, adopting a comprehensive, analytical approach should quickly become best practice. Consider developing an AI governance strategy to guarantee that your applications reflect your company values. In practical terms, this might involve ensuring that your use of AI will safeguard customer and company data, mitigate possible biases and champion transparency and accountability.

More broadly, this approach will also mean looking at and correlating a variety of key business and operational measurements, such as operational total cost of ownership, employee productivity, experience level agreements (XLAs), HR data, procurement data, customer surveys and feedback and emissions and ESG reporting, to gain a holistic view of the technology’s real-time effects across various business objectives.

In this context, the SEE pivot becomes particularly valuable. By focusing first and foremost on the environmental impacts of generative AI, there’s a chance of overlooking the broader benefits and risks this technology can bring. Envision your generative AI strategy instead as a circle of value. By understanding the nuanced interplay between customer experience, social impact on communities (and the effect that has on your brand), ESG risks and business growth opportunities, your team will be better equipped to stay on track and avoid major missteps.

Chris Kirkpatrick is a Director of Offering Management in Kyndryl's Global Digital Workplace Practice

¹ Making an image with generative AI uses as much energy as charging your phone. MIT Technology Review. December 2023.
² ChatGPT Uses a Lot of Water to Train Itself, Study Shows. Business Insider. April 2023.
³ Survey: ESG Strategies Rank High with Gen Z, Millennials. SHRM. March 2023.
⁴ Sustainable economy: Parliament adopts new reporting rules for multinationals. European Parliament. November 2022.