AI can revolutionize the response to climate change
Charlotte Degot17:272022
Most companies guess their carbon footprint, with errors of 30 to 50%
Imagine trying to cut spending while shopping in a store with no price tags. That, says BCG managing director Charlotte Degot, founder of CO2 AI, is how most companies tackle climate change: fewer than 10% measure their emissions comprehensively, according to a BCG survey of more than 1,000 firms. With emissions due to halve by 2030, the question is how companies can see what they cannot measure.
See how AI turns rough carbon estimates into decisions that can cut emissions 30 to 40%
Charlotte Degot, founder and global leader of CO2 AI by BCG, argues that companies cannot reduce emissions they cannot see, and that inaccurate carbon measurement is the first roadblock to corporate climate action. She shows how AI can estimate emissions granularly across the value chain, using a wine and spirits glass example, and calls on corporations to adopt AI now to measure, decide and cut emissions faster.
“We're almost blind about our climate emissions.”
“But the good news is it's typically the area and the type of playgrounds where artificial intelligence can help a lot because we cannot measure easily, but we can estimate a lot more easily.”
What you will learn
Explain why inaccurate measurement is the first roadblock to corporate emission reduction.
Identify where AI adds value across measuring, target setting, enabling reduction and reducing emissions.
Recognize how single average emission factors hide decision-relevant drivers such as material color, supplier country and design.
Plan supplier engagement for the indirect emissions that make up most of a company's footprint.
Weigh AI's benefits for climate action against its limitations and its own emissions.
Key takeaways from the speaker
Corporations have a major role to play in meeting the climate challenge, but few have started to address the impact their emissions have on the climate.
Lacking accurate measurement, companies set emission-reduction targets and action plans based on highly incomplete and inaccurate estimations, with margins of error of between 30-50%.
AI can help companies get to the next level of measurement and, once there, support them throughout the overall journey to emissions reduction.
AI excels in its ability to measure emissions exhaustively, accurately, and frequently, set targets and identify the best initiatives for emission reductions, and ensure delivery in overall organizations and stakeholders.
Details
Level
Intermediate
Format
Talk
Technical depth
Introductory technical
For
Executive, People manager, Senior individual contributor
Length
17:27
Published
July 2022
Questions this talk answers
Why do companies struggle to reduce their carbon emissions?
Because they cannot measure them accurately: emissions are invisible gases, and most come from suppliers outside direct operations, so decisions rest on rough estimates.
How can AI help measure carbon emissions?
AI can automate data retrieval, clean and match data, infer emissions along the value chain and extrapolate missing data, estimating what cannot be measured with sensors.
What does granular emissions modelling reveal in practice?
In a wine and spirits example, glass color, supplier country and bottle design changed emissions by factors of 1.5 to 2, differences invisible in an average calculation.
What results can AI-based carbon management deliver?
As the speaker cited from CO2 AI work, baselines adjusted by 30 to 50% versus traditional methods, and carbon reduction of around 30 to 40% was enabled.
How can awareness of inaccurate carbon measurement be raised?
By first educating about the problem; solution providers, NGOs such as CDP and regulators all have a role in talking about measurement quality.
About this talk
The climate challenge for next decade is immense. To meet Paris Agreement goals and avoid increasingly harmful effects on the climate, global carbon emissions must be reduced by 50% in the next decade. Corporations have a major role to play in meeting this challenge, but few have started to address the impact their emissions have on the climate. As the world has learned, it is exceedingly difficult to reduce carbon emissions—and particularly more so if companies have neither any idea how much carbon they actually emit nor any way to accurately measure emissions.
The problem is that greenhouse gases are colorless, scentless, and invisible, which makes them very difficult to measure. Lacking accurate measurement, companies set emission-reduction targets and action plans based on highly incomplete and inaccurate estimations, with margins of error of between 30-50%. To put the challenge simply: If you don’t know the distance between Point A and Point B, how can you estimate the time it will take to get there?
While the climate situation is dire there are, fortunately, technologies and tools that can accelerate the journey toward lower emissions. One of the most powerful tools is artificial intelligence. AI can help companies get to the next level of measurement and, once there, support them throughout the overall journey to emissions reduction. Artificial intelligence is becoming essential technology in the fight against climate change because of its ability to do things very quickly and very accurately.
Shopping without price tags: why carbon goes unmanaged (87 s)
How a bottle's color and origin change its emissions (48 s)
Eight years to halve emissions: no reason to wait (57 s)
So thank you very much for joining the session. I'm Charlotte Degault. I'm a managing director and partner at BCG. I focus on sustainability and AI and I lead, I'm the founder and global leader of a solution that is called CO2 AI. I'm going to talk to you today about how AI can revolutionize climate change or at least how it can help really get quicker and more efficient at answering to the climate change challenges.
To start with and not to lose time, I just want to start with something that is not going to be new to any one of you, which is the current trajectory of our CO2 emission and where we should be going. The red line is where we are going. And the green line should be where we should be heading if we want to limit the impact of climate change.
So, of course, needless to say that we are not on the good path and what matters to me today is to start to explore with you why and what we can do and taking especially the angle of corporations and how they can better manage their own path. In life, in general, there is a principle that applies that is quite true across everything, which is what gets measured, gets done. I would like to just stop on this sentence because this is where climate change is a real struggle and I will explain you why in a second, but it all starts from there.
When you can measure something, it's of course a lot easier to take action on it and reduce or increase it. The parallel and the image I like to take, and we could take many other ones, is just to imagine yourself trying to save money and really trying to reduce your spending. But when you go shopping, you're blind, there is no price tag on any item. This is just about being bound to fail, right? And this is close to what we have today when it comes to climate change. We're almost blind about our climate emissions.
And there are many reasons for it. It's not, it's really not easy to measure CO2 emission. It has no color, no smell, no weight. But the consequences, we don't measure enough and therefore we don't act enough because what gets measured gets done and what does not get measured does not get done. If we go in the details of it, just to share a few figures that we have been investigating with BCG over more than 1000 companies across the globe.
There is less than 10% of companies today that are measuring their emissions comprehensively. From this 10% obviously, when you don't measure, it's very hard to make decisions and to identify where to act. And only 5% of those companies are able to make and have made reduction targets regarding their CO2 emissions. And of course, when you go down the funnel and you look at how many corporations have really been able to act and reduce in line with their ambition, their emissions, it's less than 10%, less than 1%. Sorry.
So obviously we are, and we understand here, we are not on the right path to reduce climate impact and CO2 emission. And it all starts from the measure piece. If we can fix the measure piece, it will be far easier to then decide, reduce track progress. And the question is, why is it so hard to measure? And it's a data issue. In fact, as I was saying, there is no way to easily measure emissions because it's, it's, it's, it's, it's, it's gases.
It has no, no real, you should put sensors everywhere if you wanted to have accurate emissions and impossible to put sensors in everywhere. But the good news is it's typically the area and the type of playgrounds where artificial intelligence can help a lot because we cannot measure easily, but we can estimate a lot more easily. And let me deep dive into what can artificial intelligence do to help support the end to end carbon journey.
There are different aspects of things. You can look at measure, you can look at set target, enable reduction and reduce. And all along the journey, artificial intelligence can help on the measure part. Um, as I was saying, it's very hard to cover data because, and especially for large corporations, the data comes about carbon emissions come from everywhere.
And mostly on average comes from, uh, outside the direct company operations, what we call the scope free, the indirect emissions that come typically from the suppliers, uh, and the products you buy from suppliers. And things like that being able to automate data retrieval, clean data, match data, make inference about emissions produced about along the value chain, extrapolate missing data.
All this is extremely useful when it comes to climate, uh, because again, the direct data is not available so easily. So, again, the direct data is not available. So, again, the direct data is not available in the value chain because when you make a decision about climate, even the simplest decisions like logistic ones or things like that, that will come naturally to your mind.
So, again, it has many implications and being able to measure this impact accurately is can be a nightmare without artificial intelligence. It can also help create optimized roadmap and optimize not only for CO2, but also CO2 and cost or CO2 and over KPIs, waste, water, etc. Along the journey, it can also help master a lot of data to be able to track and report the progress in real time and help move to action and enable reduction a lot faster.
This also goes with optimization techniques that are very useful when it comes to directly improving the day to day decision making in operations and enable direct emissions reduction thanks to efficiency. You can think of logistics optimization, process control optimization, things like that, that you can put in place in plants or like in cement or steel, and that can directly impact the carbon emissions of big emitters.
With CO2AI, we've created a software that helps corporations master their end-to-end net zero journey. So, what we do with CO2AI is really help corporations measure accurately with artificial intelligence. And I will give you an example in a second, but also decide, simulate any kind of decisions and understand what are the best decisions to be made across the organization to optimize the journey to net zero.
And then reduce those emissions within the organization, but also engaging with the ecosystem. And the ecosystem in climate change, they are very important because 80, 90% on average of emissions come from outside the direct emissions from the company, rather from the suppliers and the suppliers of the suppliers, etc. So, it's very important also to be able to engage the full ecosystem.
And CO2AI is a software that does that. I'm telling you all this because CO2AI leverages artificial intelligence at all steps to really make the journey faster and more efficient for corporations. Let me tell you one example of how artificial intelligence works and how it can help. Measuring emissions. Let's take the example of a wine and spirits company.
A wine and spirits company typically will have a lot of its emissions coming from the glass that they buy for their bottles. Because glass is a significant bucket of emissions. And usually, because it's so complex to measure emissions, what a company would start by doing is take a total amount of volume of glass bought for a year and use one emission factor.
That is the conversion factor to be able to say, if I buy one ton of glass, this is the amount of emissions I will generate. This is obviously a good way to start and to have a first proxy of what the glass emissions represent. But it's not enough to really be accurate. With artificial intelligence, you can model and simulate all the aspects of glass that will drive emissions up or down.
Typically, the glass color, the percentage of recycled content, the country of origin of the supplier, the transport mode. And this is very significant. And this is what we've been doing with some of the users of CO2II. And it significantly impacts the total number of emissions that comes out of the calculation. But more importantly, it helps make decisions.
For example, once you've calculated your emissions on glass with the detailed calculations that I was showing, then you can simulate changes. And you can understand, for example, that the glass color is super important. There is a 1.5 factor between a flint and a green bottle. It's also the case of the country of origin. Same bottle, same production.
It can be two times more emitting in one country versus another one. It's the same for the design. You can take two different designs with the same weight and you can have a 1.5 times factor in terms of carbon emissions ultimately. This is impossible to notice. If you stay at a very high level calculation, average one, just considering the total amount of glass that you consume a year.
If you go in the details here, you can find very important decision-making areas where you can drive your emissions down by making the right operational decisions. This is why CO2 AI does, and this is what is very useful, leveraging AI to be faster and more efficient in identifying the right decisions that you're making as a corporation to drive your emissions down.
Ultimately, this is the results we achieve with CO2 AI, but this is what I'm convinced artificial intelligence can really unlock now in terms of benefits. There is a first benefit, which is the accuracy and the exhaustivity of the baselining. It's 30 to 50% adjustment on average versus traditional methods. When you do average calculations versus the granular and accurate ones, this is the results we've achieved on the glass example that I was showing with the client we were working with.
But it's something that we've been achieving across the board, working with many different industries, many different corporations across not only consumer goods, but also retail, heavy industry, pharma, healthcare, chemicals, and many others. So that's the first one, but there is a more important one that is closer to my heart, which is really how much can we achieve in terms of enabling fast carbon reduction.
And the order of magnitude is 30 to 40%. Leveraging this ability to get a true visibility on carbon emissions, and then the help to drive proper decision-making where it's the most easy to implement and efficient. If I step back, there is three things that, for me, are very important when it comes to climate and artificial intelligence. First, we have eight years to have our emissions by two.
We need to cut our emissions by 50% by 2030 if we want to stay in the Paris Agreement trajectories. So there is no reason to wait. And everything that can help on this journey is really a necessity. And we need to act now. We need also to acknowledge that the key roadblock today that is the first step to any reduction trajectory is the measurement.
It's the key roadblock for 90% of companies. And we see here that artificial intelligence can help. And it's here, it's techniques that are mastered, that are easy to implement, that does not take years to be implemented. So there is absolutely no reason not to be using artificial intelligence more in the climate field. And this is what I really want to claim.
I'm not saying artificial intelligence is the solution to everything and we should only rely on artificial intelligence. I'm also very conscious of the limitations of artificial intelligence and also the fact that we should also master the emissions related to artificial intelligence. But artificial intelligence is a solution that exists, that can help, that has proven its results. So there is absolutely no reason to wait.
And it's really a call to action that I'm making today, trying to democratize this and trying to really push corporations to act faster and more efficiently on their climate action, leveraging artificial intelligence. So let me pause here and I can take any questions that you have. Thank you, Elizabeth. It's a very, it's a very good question. So I think there is two things.
There is an awareness that is needed to be built about first the inaccuracy of the carbon measurement today. In the, in the survey that I was mentioning at the beginning, where we surveyed more than 1000 corporations, we also asked the companies about what is their estimated error rate, the margin of error they have in their current measurement of emissions, their current baseline footprints, etc.
They said on average, 30 to 50%. And the thing is, this figure is not known today. Nobody talks about it. Everybody talks about the necessity to, to reduce, but nobody really talks about the level of inaccuracy and the tool that exists. And I do think that there is a journey that is started. We're partnering with CDP, for example, which is the climate disclosure project.
And that's an NGO that is really working to push corporations to report their progress and also to do it more and more with quality data, etc. But everybody, regulation should enter into the game and everybody should, should speak about the issue of not being a, not measuring well enough and the solutions. And it can be the solution providers as us, but it should also be all types of parties and the NGOs, the regulation, they have a role to play in that.
And I think it's, it's really, before starting by the solution, I think there is education about the problem and the fact that today, really, we are not measuring accurately our emissions at all. And for me, that was the biggest, a big surprise when I started to really deep dive into carbon accounting and it was also a true discovery that nobody talks about it.
If you look on internet about it, you won't find much. That's crazy. I think. So I think we are done. I don't see any other questions. So thank you very much. It was a pleasure. I hope you enjoyed the session as much as I did. And feel free to connect with me on LinkedIn. If you want to discuss more, I would be, I would be pleased to have a nice day.
Transcribed automatically; names checked against the speaker record. Select a time to jump there.
Article based on this talk
Why companies cannot cut emissions they cannot measure, and how AI closes the gap
Charlotte Degot of BCG argues that accurate carbon measurement is the missing first step in corporate climate action, and that AI can deliver it now.
In short
Charlotte Degot, founder and global leader of CO2 AI by BCG, argues that corporate climate action stalls at the very first step: most companies do not know how much carbon they emit. Because greenhouse gases cannot practically be measured everywhere, she makes the case that AI, which can estimate emissions granularly across the value chain, is a proven tool that companies should use now to measure, decide and reduce faster.
Key takeaways
Corporations have a major role to play in meeting the climate challenge, but few have started to address the impact their emissions have on the climate.
Lacking accurate measurement, companies set emission-reduction targets and action plans based on highly incomplete and inaccurate estimations, with margins of error of between 30 and 50%.
AI can help companies get to the next level of measurement and, once there, support them throughout the overall journey to emissions reduction.
AI excels in its ability to measure emissions exhaustively, accurately and frequently, set targets and identify the best initiatives for emission reductions, and ensure delivery across organizations and stakeholders.
What gets measured gets done
Degot opens with a chart contrasting the current trajectory of CO2 emissions with the path needed to limit climate change, and states plainly that the world is not on track. Her lens is the corporation, and her starting point is a management principle she believes applies almost everywhere: "In life, in general, there is a principle that applies that is quite true across everything, which is what gets measured, gets done." (1:09)
To make the point concrete, she asks the audience to imagine trying to save money while shopping in a store where no item has a price tag. Success would be nearly impossible, and she argues that this is close to how companies approach climate today: "We're almost blind about our climate emissions." (1:54)
Why measurement is a data problem
Carbon is hard to measure for a simple physical reason: "It has no color, no smell, no weight." (2:36) Accurate measurement would require sensors everywhere, which is impossible. The consequences show up in figures from a BCG investigation of more than 1,000 companies, as the speaker cited: fewer than 10% measure their emissions comprehensively, only a small share have set reduction targets, and fewer than 1% have reduced emissions in line with their ambition.
For Degot, the whole funnel breaks at the first step, and fixing measurement would make deciding, reducing and tracking progress far easier. She frames the challenge as a data issue, which is exactly where AI is strong: "But the good news is it's typically the area and the type of playgrounds where artificial intelligence can help a lot because we cannot measure easily, but we can estimate a lot more easily." (4:23)
AI across the carbon journey
Degot breaks the journey into measuring, setting targets, enabling reduction and reducing. In measurement, AI can automate data retrieval, clean and match data, infer emissions along the value chain and extrapolate missing data. This matters because, for large corporations, most emissions sit outside direct operations, in the indirect emissions that come from suppliers and purchased products. Beyond measurement, AI can build roadmaps optimized for CO2 alongside cost or other indicators such as waste and water, track progress in real time, and support operational optimization such as logistics and process control in heavy emitters like cement and steel.
She describes CO2 AI as software that helps corporations measure accurately, simulate decisions, reduce emissions within the organization and engage the wider ecosystem. Since, as she cited, 80 to 90% of emissions on average come from suppliers and their suppliers, "So, it's very important also to be able to engage the full ecosystem." (7:43)
What a glass bottle reveals
The most memorable part of the talk is a worked example from a wine and spirits company, where glass bottles are a significant source of emissions. A typical first approach multiplies the total tonnage of glass bought in a year by a single emission factor. Degot calls this a good first proxy, but not accurate enough. Modelling the drivers instead, such as glass color, recycled content, supplier country and transport mode, changes the total and, more importantly, reveals decisions.
"There is a 1.5 factor between a flint and a green bottle." (9:48) The same bottle can be twice as emitting depending on the country where it is produced, and two designs of equal weight can differ by a factor of 1.5. None of this is visible in an average calculation. As the speaker cited from CO2 AI work across consumer goods, retail, heavy industry, pharma, healthcare and chemicals, granular baselines differ by 30 to 50% from traditional methods, and the order of magnitude of carbon reduction enabled is 30 to 40%.
A call to act now
Degot closes with three points: emissions must be cut by 50% by 2030 to stay on the Paris Agreement trajectory; measurement is the key roadblock for 90% of companies; and the AI techniques that help are mastered and quick to implement. She is careful not to overclaim: "I'm not saying artificial intelligence is the solution to everything and we should only rely on artificial intelligence." (13:46) She also acknowledges AI's limitations and its own emissions, yet insists the tool is ready and proven.
In the Q&A, she returns to the survey: companies themselves estimated a margin of error of 30 to 50% in their current footprints, a figure she says is rarely discussed. "Everybody talks about the necessity to, to reduce, but nobody really talks about the level of inaccuracy and the tool that exists." (15:32) She sees a role for solution providers, NGOs such as CDP and regulators in educating about the problem before selling solutions.
Watch the talk
"AI can revolutionize the response to climate change" by Charlotte Degot, about 17 minutes: https://www.womentech.net/video/ai-can-revolutionize-response-climate-change-charlotte-degot
About the speaker
Charlotte Degot is Founder and Global Leader of CO2 AI by BCG at BCG - Boston Consulting Group, an AI-powered solution that helps corporations advance on their sustainability journey. She is also a Managing Director and Partner of BCG GAMMA, BCG's data science and advanced analytics entity, and the firm's global climate and sustainability lead.
Questions this talk answers
Why do companies struggle to reduce their carbon emissions?
How can AI help measure carbon emissions?
What does granular emissions modelling reveal in practice?
What results can AI-based carbon management deliver?
How can awareness of inaccurate carbon measurement be raised?
Drafted with AI from the talk transcript. Editor review pending.
Frameworks and ideas mentioned
Paris Agreement
Scope 3 emissions
Net zero
What gets measured gets done
Measure, set targets, enable reduction, reduce
Cite this talk
Charlotte Degot. "AI can revolutionize the response to climate change." WomenTech Network, 2022. https://www.womentech.net/video/ai-can-revolutionize-response-climate-change-charlotte-degot