EP 38: How AI is Transforming Sustainability

In this episode, we sit down with Rochelle March, Founder of People Places Words Action (PPWA) to discuss how AI is reshaping the sustainability landscape. Rochelle shares how companies are using AI to boost efficiency, uncover insights, and tackle complex global challenges like climate risk, biodiversity, and supply chain impact. She also explores the balance between humans and AI, the rise of responsible AI practices, and how sustainability professionals can build their own AI capacity to drive change.

About Sustainable Intelligence 

Novata’s Sustainable Intelligence podcast brings honest, forward-thinking conversations about sustainability in the private markets. From carbon accounting to turning data into impact, each episode explores how investors and business leaders are tackling real challenges, meeting rising regulations, and building long-term value. Join us for expert insights, practical strategies, and authentic dialogue that empower action and progress in the evolving sustainability landscape. Listen to more episodes on Spotify or Apple Podcasts.

Conversation with Rochelle March (Transcript)

 Ella Williamson: Welcome to Sustainable Intelligence, where we discuss all things sustainability for the private markets. I’m Ella Williamson, and I’m thrilled to be your host. Joining us today is Rochelle March, Founder of sustainability consultancy, People Places Words Actions (PPWA). She has over a decade of experience in ESG, data science, and impact strategy. Plus, she’s helped shape how companies measure and manage sustainability worldwide. Today we’re going to explore how AI is transforming sustainability, from boosting efficiency and uncovering new insights to empowering human-led innovation. Rochelle, so great to have you here. To get us started, I’d love you to tell us a bit about your incredible journey and your career in sustainability.

Rochelle March: Thank you, Ella. It’s a pleasure to be here. I started out my career out of grad school within bespoke consulting firm called Sustain Ability. There, I really got a diverse perspective of different customers, different segments, different projects, ranging from competitors, regulation, innovation opportunities, investor relations. And I later joined S&P Global in their ESG innovation analytics division, where I really cut my teeth on the technical applications of data analytics models and leader product development.

And so when I was offered an opportunity to build out small, medium business, private company, and supply chain information around ESG at Dun & Bradstreet, I accepted that opportunity and built out their ESG division and product offerings. And this was a very rewarding experience. And once I was successful in that endeavor and we launched a revenue generating product suite in data suite, I would get similar offers. Could you come over here and help me build an ESG business? I felt at that time, I might as well build my own. That’s when I started People Places Words Actions with my business partner where we hope to service mostly individuals in organizations who are trying to accelerate projects or make sustainability oriented changes.

Ella: Brilliant. A very exciting and impressive career you’ve had. And with many other spaces, AI is the buzzword that’s on everyone’s lips. From your experience, how are companies starting to use AI within their sustainability strategies, and are there any interesting ways you’ve seen them starting to integrate it into their daily workflows, et cetera?

Rochelle: Sure. So this is obviously a major focus of many organizations. How do we integrate and adopt AI? And a helpful way of thinking about this is a framework that we’ve put out recently, breaking up organizational readiness when it comes to AI. So there’s three pillars: support, augment, and discover. And the first one, support, is really about how do you essentially boost human productivity through AI?

So we’re thinking about efficiency gains, automation, maybe extracting a bunch of data points from a report. I’m thinking of Perplexity AI’s SEC filing search when you need to learn about different companies and what they’re doing and their financials or risk factors and their annual reports. And this is really just speeding things up, like making your job easier and your workflows faster. And this is predominantly where most organizations are focused on today. How can they increase and accelerate their employee’s work and work streams?

The next one, augment, is where you are able to leverage AI to take on more advanced skill sets and activities. So an example might be a chief strategy officer or sustainability officer using AI-generated scenario models to assess risks and trends. Scenario models and planning is a multi-varied domain endeavor that often requires weeks and weeks of iteration. But with AI tools, you can really take on that higher, advanced, complex activity.

Another example more in the medicine and academia route is work being done at the University of Pennsylvania, where they’re using an AI-generated database to identify existing drugs that have already been FDA approved in the US, for example, that can stabilize and treat patients with rare diseases. And so these patients generally never get the type of medicine they need  ’cause rare diseases are rare. There’s not a lot of funding allocated towards them, but if there’s an existing medicine they can use to treat their disease all the better, and that’s really amazing, right? Like that’s expanding the ability to to treat these patients and for those doctors to recommend different things that otherwise we’d have to go through many lab trials and and sequences.

And then the final one, discover, is really where you have human-led intuition guiding AI, where AI is doing a lot of the brute force and calculations or organization and humans are directing it where it needs to go based on our broader understanding. And this is predominantly in academia and science. There’s some companies like Matlantis, for example, that train AI on quantum mechanics to predict material properties. And so this is super helpful in terms of maybe accelerating like new batteries, solar, or catalyst designs. I think they claim to be able to do that 10,000 times faster than traditional methods. But this is more still academia and science and really amazing breakthroughs are happening here in the scope of what’s called foundation models. And I think more and more organizations will get to that place in the next couple of years. And we may see some really amazing opportunities and discoveries in that realm.

Ella: Fascinating, and really useful to hang that off those three pillars. I guess it depends on the level of adoption and where they are in their AI adoption journey as to where they are in that framework. Now, you’ve touched upon some really fabulous examples there. Are there any particular examples that you’ve seen and worked on where you’ve actually seen AI improve sustainability results across the company?

Rochelle: Yes. I, for example, am working with an ingredient company that services a lot of the major CPG brands like Danone and Unilever and so on. And they basically build ingredients. And so now they’re using AI to build lots of different information, so not just what’s the mouth feel and the taste and how does it dissolve and water and so on? But also what’s the carbon footprint? What’s the water impact? The biodiversity, and putting that all together and then being able to formulate new ingredients that hit the product requirements, but also have a better sustainable impact. So I’ve been working with that company for a while to basically get to a point where we could even build the database that has all of this information and then build predictive analytics using AI to help with that ingredient formulation. So that’s really exciting.

I also work with a firmographic intelligence company that builds data on all companies around the world, very AI native tech stack and everything that they’re doing with custom LLMs. And this is so critical in terms of shining sustainability information on private entities worldwide. Really granular information like all their locations that can be applicable to carbon emission calculations, biodiversity, climate risk, understanding supply chains. And so it was very exciting for me to work with these companies and just to see how much can be done and uncovered with AI and how quickly to really meet more of the requirements of organizations today that need real-time information, highly customized to customer needs and also reliable metrics that can be explained with some of these.  

But, other than my own portfolio, there’s a number of different examples I can mention that are neat in terms of how AI is used. Unilever for example, I was reading, uses AI and integrates it with weather and sales data to improve demand forecasting. A cute example is in Sweden, this enhanced seasonal product forecast like ice cream by 10%. So if it’s very cloudy and cold, people are less likely to buy ice cream. And this is good from a sales perspective, but also from their cold chain, that is very energy intensive and if you are able to decrease that based on demand, that has a sustainability benefit.

Then, another example I’ll mention is more related to agriculture. So John Deere has an AI-powered sea and spray system that identifies the plants, identifies what type of herbicide and how much they would need. And reduced herbicide use by up to 59%, which not only improved the yields of these crops ’cause they got what they needed to grow well, but also protected soil health and reduced the amount of herbicide you’re using there. So, these are just a couple of examples of how AI, kind of, optimizes systems to make them more efficient and thus more sustainable, as well as discover new ways to complete those business operations in the field.

Ella: Absolutely. And I think at first glance you may think that actually predicting what the weather’s gonna be in your sales forecast isn’t directly related to sustainability at all. However, the energy that would then save is massive. So, a really great comparison there. Now, I know that a lot of people worry that AI could replace their jobs whereas others really see the benefits of it, like you outlined in those examples. How do you see the balance between humans and AI evolving over the coming months and years, and are there any trends you are noticing in terms of hiring, skill sets, and how people use AI on a day-to-day basis?

Rochelle: Definitely. So, in many ways we’re back where we were in say 1993 when the worldwide web emerged and the global economies transition to a digital first type of transaction and interactions. And very much in that vein, we are experiencing the introduction of new technology and arguably being implemented much faster than the worldwide web. So there’s gonna be shifts and transitions a hundred percent. And we are already seeing that, right? AI is contributing somewhat to a lack of job growth along with other factors globally. I was reading an article how over a hundred thousand tech-specific jobs have been let go due to AI and what’s happening. Unemployment for new grads is at a pretty big high and so on, so this is affecting us.

But it is up to us and new grads and employees to use AI to maximize our own personal opportunity as well as sustainability impact. In organizations, maybe you had 20 people to analyze reports and understand ESG performance or put together campaigns to push out new ideas. And in the future we may have much less, we may have five people who do that, who are able to support the workflow through AI and what have you, and because of that, I think it is important to not only upskill on AI and see how I can enhance and amplify your own skill sets and abilities, as well as learn new skills and abilities, build a portfolio of skills, but also adopt this entrepreneur and proactive mindset in one’s career and within organizations where we can learn new things very quickly. We can take on more complex projects, we can build automations, we can build better communication systems across teams and really embrace it.

And then we do see some regulation coming out to keep track of, kind of, how many jobs are being displaced because of AI. I’m thinking of the New York’s WARN Act and local law in New York City, Local Law 144.  But I think overall, it’s important to think of AI as a tool and to engage with that tool and to use it for your own endeavor, which for sustainability professionals is really important. Our work is critical and to maximize our impact is also critical through all the tools that are at our disposal.

Ella: I’ve heard a lot of conversations about AI and read a lot of articles about AI, but I’d never heard anyone look at the benefits of being that ability to open up entrepreneurial mindset, and I really like that framing.  Within the same vein, with the growth of AI and everything that’s in the news, responsible AI seems to be a term that is used very regularly. What does that really mean in sustainability practices, and especially what does it mean with things like privacy and the future of work that you just discussed?

Rochelle: Yes. This is obviously a very dynamic topic and being shaped by current lawsuits and regulation and also just worker employee sentiments. And it’s also critical that sustainability professionals are part of this conversation, right? Like, how do we use AI to be responsible, and to have the attributes that we wanted to have in a sustainable context. I think it’s important to first talk about some seminal activity that’s happening with privacy, right?

So there’s a number of regulations globally, actually, that don’t allow, say, non-consensual deep fakes, for example, or non-disclosed surveillance like in facial recognition, as well as I mentioned, some mandatory disclosure on layoffs related to AI to capture how this is affecting people’s jobs and the economies. So this is top of mind in many ways. And also the EU AI Act, now enforceable, outlines a lot of these things as well as US and some Asian regulations too. So that’s happening, which is good, and we’ll continue to see that. But regulations do tend to lag behind actual business activity. Related to actual business activity, some seminal lawsuits that are good to keep in mind in terms of privacy and how these tools can be used. Top of mind for me is The New York Times suing ChatGPT because they were worried that ChatGPT went behind the paywall and used a lot of The New York Times data to train their models. And in order to prove that, they were able to access previously deleted AI chat logs hosted in the cloud. And so this is concerning in the way that, can we really trust AI providers’ data policies? And it’s also a little bit concerning in the sense that we store a lot on the cloud.

Think of our emails or enterprise accounts and AI chats aren’t fundamentally different from that. They’re also stored in the same place. So this kind of raises new privacy and security considerations that are actively being addressed and worked out in the landscape today. But I think it’s important to think about it in terms of your own use with these tools. And it’s very unlikely that some search query that you have is going to show up in somebody else’s, just the way that these models work in a probabilistic way. But it is important to keep in mind. You don’t necessarily want to put out something you wouldn’t want people to see in the sense that technically it could be retrieved as it has for some of these lawsuits. There’s also the ethics around training of copyrighted content. So there too, when we’re putting out content, these models are taking it from the internet essentially and training the models on it. So that’s obviously a big area and also hinges on these issues of who owns what and what’s intellectual property to individuals as well as organizations. 

But when it comes to your own workplace and what you’re doing, practically speaking, if your organization has approved an enterprise LLM, like Gemini or Microsoft Copilot or Open AI, or what have you, they’re operating on the same principles as any other type of enterprise software, say email and drive and storage, and the same. So you should be totally fine to be able to use it for business purposes and to really leverage all of the capabilities these tools have. Now through that, your organization probably is going to give you some training around, never upload confidential information outside of these tools, never upload personal identifiable information, including your own, in your own personal use. And as I mentioned, as a general rule of thumb, you probably don’t want to put anything in these tools that you wouldn’t you a hundred percent wouldn’t want anybody to see. Just as a general rule of thumb, however. I caution against this being a blocker to using these tools because they are so helpful.

They are so incredible in their capabilities, and these are just important factors to keep in mind in your own work, but also in how we develop policies and ways of working in organizations when it comes to responsible AI and how we want to implement them in our own organizations as well as any products and services that we’re offering.

Ella: Thank you for that. I think, like you said, being mindful and thoughtful with how you use AI and following the rules that are available to you is the real key here. Being in sustainability, carbon footprint is something that we talk about a lot, and there’s growing awareness that AI has a pretty hefty carbon footprint.

What’s your perspective on this, and is it something that we should be concerned about? Would love to get your thoughts.

Rochelle: Sure. Certainly it’s a large footprint, it is increasing emissions. We’ve seen many companies, tech companies, for example, miss carbon reduction targets because of the increase of AI-related emissions. I was reading recently, US data energy use is projected to reach about 6% of national electricity consumption, which is significant. In comparison, however, it’s very modest, compared to some sectors like air conditioning or even expanding the electric vehicle fleet, which consumes much more energy than this. So it is an increased footprint, but it is like part of technological development and that need to power that.

Now what’s interesting from a carbon footprint perspective is there’s a number of studies that first of all, there’s a lot of activity done to make the building of the models—which is where most of the energy comes from—much more efficient, much less energy intensive, not only from a sustainability perspective, but from a cost perspective and an optimization perspective from a business point of view. So, with that, there’s the fact that AI will probably become much more efficient and data centers will as well. This happened similarly, I believe, in the late 2010s or so when data center energy was increasing tremendously and then they implemented a ton of efficiency measures and the energy kind of plateaued. So we’re back there again where we need to figure that out and build that efficiency back in. But there’s also obviously the benefit of AI to optimize and reduce energy use elsewhere. And so that adoption of AI through other sectors is critical. Also in how we can use it to manage climate risks and other components. And the International Energy Agency put out a pretty great AI and Energy Report with some scenarios and essentially from 5 to 10 years, if AI adoption occurs, we could reach carbon neutrality, not only related to AI specific use, but beyond that. So there’s a net gain predicted. And, Nature put out a recent study here too that projected an even greater greenhouse gas emission reduction overall, totally zeroing out any data center related emissions.

Not ideal, right? Ideally you have a net positive solution right off the bat. But I think if we keep going and really leverage AI to find better solutions and more efficient solutions in all sectors, including data centers, we’ll actually see a net gain related to the carbon footprint. And as a sustainability professional, you should not keep this, have this fact, the carbon footprint, be what keeps you from using these tools because your individual use of it is so tiny compared to anything. And we need all the tools we can get to achieve our goals. So don’t have this as a factor keeping you from really learning and upskilling and leveraging AI for your own impact.

The last thing I’ll say is for those practitioners who are really trying to measure the carbon footprint for AI, there’s a lot of interesting elements that have come out. So Salesforce has collaborated with Hugging Face and Coface and some other organizations to create an AI energy score. So this essentially scores mostly the open models, but some commercial models too, I believe, in terms of their energy use. Then there’s been some pretty interesting papers out there that have basically reverse engineered the energy use of proprietary models, so those that are not open source. And I believe one of the key findings was the Claude 3.7 sonnet model scored the highest in terms of eco efficiency, which is not surprising. Anthropic has led with their more responsible brand and thoughtfulness there. And then I believe Deep Seek was 70 times more energy intensive than GPT 4.1 nano, which is a relatively small model.

Not a great comparison, but just some numbers to throw out there and some examples of how there’s a lot of movement around this if you are interested in measuring specific AI usage to your organization.

Ella: Yeah, and even though AI may not be net positive in terms of its carbon footprint today, like anything, it will take time. And the benefits but also the impacts that it has, there’s positives and negatives for both. One big question that I’d love to get your thoughts on is everything’s changing so fast, everyone’s trying to keep up. What are the big shifts that you expect to see in the AI space, specifically for sustainability, over the next few months, over the next few years, whatever you think is a realistic timeframe to look at?

Rochelle: First and foremost, this is just a really exciting time and I spoke about this AI adoption framework in the beginning of our conversation, support, augment, discover. And I think we’re just gonna see more and more organizations build their AI capacity and get further along in that framework, where you’re not just making your workflows super efficient, you’re able to expand those workflows across new capabilities and more complex activities to then being able to discover whole new solutions in your field of expertise, whether that’s healthcare, food, energy, battery, what have you. So I think we’re going to see that and see really an integration of AI technology into how we develop, expand, and scale solutions, sustainability and non-sustainable alike.

In terms of sustainability, sustainability is challenging in the sense that it’s based on systems that are highly interrelated, dynamic, synergistic, and quite frankly, very challenging to model in traditional methods or to find solutions in say like traditional statistical methods or simply mapping it out and understanding all these relationships. So I do think AI not only can accelerate existing workflows but we can also start to tackle some of these really complex things when it comes to climate, ecosystems, biodiversity, global economic interactions, the footprint of those interactions, the social impact of those interactions, and what have you.

And I think that’s where I’m really excited to see how sustainability professionals really leverage this tool to tackle some of these highly complex things. So, to ground that, one example is the challenge of circular economy. With AI, you can say build a digital twin, which is basically a digital copy of a product or even a factory or even a network of factories, and you could test out different scenarios of how, let’s say this network of factories could build a circular value chain, build this product and the waste that product goes into this factory and that goes here and so on, and really start to design and understand and test out what this implementation could be and then go and implement it. And so I think that is really exciting in terms of how we can use AI to really harness this complexity and find design solutions around that and see those more sustainable outcomes.

Ella: And I think the sustainability community always stays quite close, so it’s also sharing those learnings, those solutions to really help the broader community move forward. Now Rochelle, we always finish these episodes on three quickfire questions. Whatever springs to mind would be fabulous. Number one, what’s the most common misconception you hear from clients about using AI in sustainability?

Rochelle: Yeah, so I think  there are a number of sustainability oriented challenges with AI, which they think of first. So the carbon footprint, the privacy issue, the job issue. And these are challenges, but there’s a huge opportunity to actually be part of this conversation and to shape how AI is used. And to do that, you need to learn about AI and you need to use it and gain an understanding to speak intelligibly to it. 

On a personal note, I do a number of workshops and an upcoming course at University of Vermont around applied AI for sustainable business with MBA students. And the top issue I’ve heard from them is they’re very hesitant to input or upload any of their own work into these tools because they’re afraid that it will be used for training data and what have you. And, I would just say, if it’s your novel that you’ve worked on your whole life and you’re ready to publish, maybe not, keep that to yourself and go through a publisher. But if you’re writing a draft, if you’re working on an article, these are very helpful tools in terms of doing copy editing or synthesizing or finding new information. So you know, that fear of that data being used for training data or privacy issues I would just weigh with the benefits of using these tools and again, don’t keep the footprint issue or the responsibility issue from you actually learning and using them. In order to be part of these conversations and shape how AI will be used in the future.

Ella: Next question, which sector has been the quickest to adopt AI for sustainability in your experience? And has there been a sector that you’ve seen that actually has been the slowest?

Rochelle: Sure. So I think actually healthcare has been quite a fast adopter to AI. I know personally I go to the dentist now and they have a tool that identifies all the different parts of the teeth, and the dentist simply comes in and double checks, validates essentially, and it really streamlines the time and the process of that. Radiology is another area where the pattern recognition of disease and pathogenesis diagnosis is highly streamlined through these tools. You still need a radiologist to confirm, to double check, to look at this, but it actually, by training on the expansive human knowledge of disease imagery, you’re able to get a more accurate result often which is really amazing. 

And then the second sector would be tech, I heard a stat the other day, it was, I think 80 to 90% of programmers and engineers today are using some type of AI-oriented coding tool. And I can totally see why, for those of you who’ve used AI to code or to program, you don’t have to memorize so much or look at a bunch of sites like Stack Overflow and what have you, to get that function and get the syntax correct. You can simply use these tools to give you the code and use them and validate them and test them. So those are definitely quick to adopt sectors, I would say.

And then those that are slow are, I would say, ones that have more regulation associated with them. So surprisingly, finance, banks, there’s a huge appetite to use AI, but they are challenged with the regulatory hurdles and need to make sure all of that is in coordination with what needs to happen from a regulatory perspective. And then the other sector that kind of comes to mind is actually like real estate and construction, which arguably has a lot of gains to be had from AI, but digitization in that sector is uneven. You’ve got equipment, you’ve got some things digitized, you’ve got actual materials and building that’s happening. So that’s a place too where I think we’ll see a lot of acceleration maybe later in the game.

Ella: Fascinating. Thank you so much. And last but not least, if listeners could remember just one thing about AI’s role in sustainability from today’s session or anything else in your brain, what would it be?

Rochelle: Yeah, I do a number of AI sustainability workshops for students and corporates alike. And I always start with the fact that we have very, very ambitious and huge goals ahead of us as sustainability professionals. I think it was a Microsoft report that consolidated some of the big targets we need to hit to meet the global goals like the UN SDGs. In less than five years, we must triple carbon free energy, double energy efficiency, completely reduce biodiversity loss. So these are very, very complex, tough things to do, and arguably, AI represents the most powerful tool we have to currently meet these challenges within these very tight timeframes, right? This is a big urgency, a large scale of work. So, my takeaway would be, as sustainability professionals, your work is just too important to not use every tool and skillset available to you. And to learn as much as you can to pursue your sustainable agenda. And as I mentioned earlier, to be active in the dialogue of how AI is used so it can be used responsibly. So go out there, learn more, and definitely use everything you can to push sustainability forward because it’s so critical and so needed.

Ella: Amazing. I couldn’t think of a better way to finish this, but saying that sustainability professionals, not using AI, their work is far too important for that. So utilize it in whatever way you can.  

The things that really stuck out today to me, Rochelle, was building your AI capacity using that framework that you have repeated a couple of times—support, augment, discover. Thinking about AI to enhance your entrepreneurial mindset and the ability to do more than you’ve ever been able to do before. And finally, probably building on that, starting to tackle those bigger, more complex issues that you haven’t been able to do in the past, be that due to resources or what you are capable of, so we can start to tackle those more complex issues like climate, ecosystems, and biodiversity.

Rochelle, it’s been absolutely fascinating, I’ve absolutely loved our conversation today. 

Rochelle: Thank you so much.

Ella: Until next time, let’s keep building sustainable intelligence together. Find out more at novata.com.