AI Meets Earth Day: Finding a Sustainable Path in AI-Driven Marketing
AI Meets Earth Day: Finding a Sustainable Path in AI-Driven Marketing
What marketing leaders need to know about the environmental impact of generative AI, and seven ways to align AI innovation with sustainable marketing practices.
Today marks the 54th anniversary of Earth Day, the annual reminder that our planet is worth protecting. A range of global brands — from ASICS to Xbox — are celebrating the event with marketing campaigns that highlight their own sustainability efforts and impact. And because it is 2024, after all, it’s likely the agencies and in-house teams that developed these campaigns have tapped into generative AI to spotlight insights, brainstorm concepts, or create images and copy.
Even climate activist Greta Thunberg’s Fridays For Future used AI to generate portraits of 20 world leaders as children, each holding the earth in their hands.
It’s ironic, given that generative AI can take a substantial toll on the planet — but hardly surprising. According to a roundup of sustainability stats from eMarketer, sustainability barely cracks the list of senior marketers’ top 10 priorities. And while sustainability’s importance has indeed grown over the years and will continue to rise in the coming years, its perceived value is dwarfed by “innovation and the adoption of new technologies.”
AI Might be Part of the Sustainability Solution
Granted, sustainability and technological innovation aren’t necessarily a binary choice. In fact, tech leaders like OpenAI’s Sam Altman to Nvidia’s Jensen Huang cite climate change as one wicked problem that AI can help humanity solve. Jeff Bezos believes so strongly in AI’s ability to address climate change that he’s launched a $100 million competition seeking solutions. Nvidia’s Earth-2 taps proprietary generative AI models to forecast extreme weather events. To my knowledge, OpenAI isn’t active in this area, despite Altman’s pie-in-the-sky AI solutionism — and the GPT-maker’s planned $100 billion data center (with Microsoft) is bound to be an environmental disaster on its own, as Altman urges world leaders to explore new energy sources that better support his personal drive to achieve AI dominance.
Beyond the AI hype bubble, AI researchers generally agree. Artificial intelligence could indeed help address environmental challenges — democratizing access to weather data, optimizing systems for reduced energy consumption, enhancing renewable energy integration, streamlining waste management, and more.
When they tout these benefits though, they’re likely to reference sophisticated machine learning models, predictive analytics, and complex business intelligence systems. They’re decidedly less optimistic about generative AI in particular, often pointing to power-hungry large language models (LLMs) and other GenAI systems as part of the problem, not part of the solution.
In short, popular tools like ChatGPT, Claude, Bard, Midjourney, and (when it arrives) Sora aren’t likely to deliver wins in the fight against global warming — certainly not on their own. They aren’t meant to. And in practice, they’re moving the needle further into the red.
AI is Definitely Part of the Problem (and Marketers Need to Understand the Impact)
It’s important for marketing leaders to understand the environmental impact of AI — regardless of whether environmental impact ranks on their own list of brand priorities. After all, March 2023 data from Gallup and Bentley University shows that 55% of US adults say that brands should have a public stance on climate. Ignoring the impact of your AI applications isn’t a sustainable strategy.
Training an LLM (to focus on the most popular GenAI model type) involves multiple high-power GPUs operating continuously, which not only consume large amounts of electricity but also generate considerable heat, requiring robust cooling solutions that often use significant quantities of water. The training process for such models typically spans weeks or months, involving vast amounts of data processing across many servers, further contributing to high energy and water usage.
For instance, the Microsoft data centers that supported the training of GPT-4 experienced a substantial increase in water usage, with one report indicating a 34% rise from 2021 to 2022, amounting to nearly 1.7 billion gallons of water. This surge is largely attributed to the demands of AI research and development, including the cooling requirements for the powerful computational hardware needed for training such sophisticated models.
The energy demands are also considerable. Training GPT-4 involved vast amounts of electricity, estimated at 1.287 gigawatt hours. This is roughly equivalent to the electricity consumption of 120 US households over a year. The environmental implications of these figures are significant, highlighting the resource-intensive nature of developing cutting-edge AI technologies. Efforts to mitigate these impacts are crucial, with ongoing research into more sustainable practices and the use of renewable energy sources to power these operations.
Granted, many marketers aren’t training their own LLMs from scratch, so this may seem like a sunk cost outside a brands’ (or agencies’) immediate area of accountability. But the environmental costs of generative AI usage alone can be staggering, too.
A single ChatGPT query can use up to 15 times more energy than a Google search and gulp around half a liter of water. Multiply that by an entire marketing team using GenAI tools throughout every stage of the campaign lifecycle; by everyone across your enterprise as generative AI gets baked into every business software application; and ultimately by every customer that interacts with your chatbot for access to customer service, sales, or marketing information.
If data centers account for around 2% of US energy consumption today, AI could “gobble” up to a quarter of US energy by 2030 as it surges in popularity and everyday usage. And a significant chunk of that surge will come from end-user organizations (like brand marketers and their agencies) and the consumers they serve with increasingly common AI-powered marketing touch points.
So, given that I’m writing this on Earth Day, I’d like to encourage every marketer reading this post to consider their personal accountability where AI intersects with the environment.
How Marketers Can Bridge Technology and Sustainability: AI for People, Planet, and Profit
Since CognitivePath is an AI consultancy, we’d hardly advocate that brands sit out the AI revolution. But we certainly urge enterprise marketing leaders to think through the implications their AI decisions have on the future of people and the planet (in addition to profit) — and act with AI in ways that serve consumers, build trust, and benefit everyone.
With this in mind, marketers who want to integrate AI into their strategies while maintaining a focus on sustainability should consider the following.
Educate the Organization on Efficient AI Usage: It's essential for enterprises to use AI technologies efficiently to minimize their environmental impact. For marketers (and across the entire enterprise), this includes optimizing the use of AI applications to avoid unnecessary computational tasks, effectively managing data, and leveraging AI to improve overall business efficiency, such as reducing energy use or material waste in other parts of the business. Don’t assume that employees understand the environmental impact of simple, everyday generative AI usage — the routine running of prompts for fact-finding, brainstorming, copywriting, summarization, etc. Educate your workforce so they can make informed choices about where the benefits of GenAI outweigh the costs. Savvy organizations will bake this into AI upskilling and capture the corporate point of view as part of AI usage policies and guardrails.
Balancing Efficiency and Ethics: Marketers must weigh the efficiency gains from using AI against the potential environmental impact. For example, AI can optimize digital ad placements, improving campaign performance and reducing wasted impressions, but the computational resources required to run these AI models may be substantial. Part of the answer may lie with next-generation programmatic advertising marketplaces like TRUSTX, a Certified B-Corp ad network that minimizes environmental impact by focusing on clean, premium inventory. Ultimately, the decision to use AI should consider whether the improvements in marketing effectiveness justify the environmental costs.
Sustainable Data Management: Data is central to AI-driven marketing, from consumer behavior analytics to personalization strategies. Marketers need to adopt sustainable data management practices, such as minimizing data wastage by maintaining high-quality, relevant data and retiring redundant or outdated data. Efficient data management can reduce the energy consumption of data centers that support AI systems, aligning with sustainability goals.
Choosing the Right Tools and Partners: Marketers should select AI solutions and partners that prioritize sustainability. This includes choosing vendors who use energy-efficient data centers powered by renewable energy and who design AI models that are resource-efficient. At a minimum, marketers should understand the commitments their Big Tech partners have made to minimize the environmental impact of their technologies and operations. Google and Microsoft have public positions on technology and sustainability (even if their real-world impact isn’t always clear), while Meta provides greater transparency into the energy impact of open-source models like LLAMA-3 (the training of which required substantial resources, but that Meta claims to have entirely offset as part of its wider sustainability strategy). Beyond this, marketers should consider alternatives to LLMs, especially as the use of more resource, data, and energy efficient approaches — like small language models — become more practical for more business use cases.
Transparency and Consumer Trust: There’s growing consumer demand for environmentally responsible practices. Marketers using AI should be transparent about how they use AI and its impact on sustainability. Reporting on a given AI application’s carbon footprint, water usage, and energy consumption should be part of broader sustainability reports. This transparency can help build trust with consumers who are increasingly making purchasing decisions based on corporate sustainability credentials. At the same time, brands must steer clear of both greenwashing and “AI washing” in both their communications and their products.
Regulatory Compliance: As regulations around both AI and sustainability tighten globally, marketers must make sure that their AI implementations follow these evolving standards. This includes adherence to data protection laws, which can influence how AI processes personal data (a vital consideration for big-picture sustainability efforts that emphasize people in addition to planet), and environmental regulations, which may dictate how data centers and AI operations minimize their carbon footprint.
Innovation in Sustainable Practices: Finally, marketers have the opportunity to use AI not just for enhancing marketing efficiencies but also for innovating new sustainable practices. For example, AI can help in developing more sustainable products, optimizing supply chains for lower carbon footprints, or even aiding in the circular economy by predicting product life cycles and recycling needs. As we pointed out during our presentation at the Association of National Advertisers’ inaugural AI for Marketers conference, AI — even Generative AI — is good for much more than just creative automation and personal productivity. It can play an instrumental role in everything from customer understanding and brand strategy, to product development, smarter distribution, demand forecasting.
By carefully considering these implications, marketers can effectively leverage AI to not only enhance, accelerate, and innovate their strategies but also uphold and advance their commitment to sustainability.
How are you aligning AI innovation with sustainable marketing practices on Earth Day — and every day?