In a recent article from the World Economic Forum the authors made the point that artificial intelligence (AI) systems will be essential for the emissions mitigation task that society faces. In fact they go so far as to say that “without AI, we won’t meet ESG goals and address climate change”. In the article they refer to specific systems now coming into use, with different systems tasked with managing particular tasks within the overall mitigation journey. They conclude by noting;
Corporations that connect all their relevant data within a unified AI platform — not just ESG and financial data streams — will benefit the most. When corporations can see the entire picture and understand all the ESG metrics, only then will we produce goals that provide meaningful change and avoid the worst effects of climate change.
At the societal level, AI platforms are also starting to emerge and they are being designed to tackle a broad range of issues and questions by being introduced to huge amounts of data, such as you and I might come across if we sit all day and trawl through hundreds of internet sites and pages. The skill of the AI platform will be to integrate this data and use it to derive clear answers to questions and offer proposals for change so that progress can be made. Ideally the systems should develop insights that are not obvious to humans looking at the same data, perhaps because of the amount of data presented, which the AI system can easily handle.
Late last year one of the newest versions of such a system was made available for public testing. It is called ChatGPT and can be accessed here. To be fair this is a system that is designed more to engage in a conversation than to solve the world’s problems, but it does still draw on an immense data resource in compiling its responses. As such, I thought it would be interesting to engage with the system over the Christmas break and see what conclusions it might draw from a discussion on emissions mitigation. Note that the answers below are abbreviated, simply to limit the length of the blog post.
I started with posing a simple question;
Q: Atmospheric CO2 is now over 420 ppm. Why is this a problem?
A: Rising temperatures, changing climate, ocean acidification and therefore overall negative impacts on human health. Overall, it is important to reduce our greenhouse gas emissions and limit the concentration of CO2 in the atmosphere in order to mitigate the negative impacts of climate change and protect the health of people and ecosystems.
Q: So how might carbon dioxide emissions be reduced very quickly, say halved by 2030?
A: Increasing the use of renewable energy, improving energy efficiency, planting trees, electrifying transportation and making use of carbon capture and storage: Carbon capture and storage (CCS) is a technology that captures CO2 emissions from power plants and other industrial facilities and stores them underground. CCS has the potential to significantly reduce CO2 emissions from these sources, but it is not yet widely deployed.
So ChatGPT certainly understands the issue and is clear on the direction of travel. It also sees the importance of CCS without any prior reference to the technology by me (other than it may have read my blog posts as part of its learning), so that’s a good start. But getting it to be specific on how quickly these technologies could be scaled and how much difference they might make by 2030 turned out to be much more difficult.
Q: How quickly would these steps need to be scaled to cut emissions in half by 2030?
A: The rate at which these measures would need to be scaled up in order to cut carbon dioxide (CO2) emissions in half by 2030 would depend on the current level of emissions and the specific measures being implemented. For example, transitioning to electric vehicles (EVs) and public transportation could make a significant dent in transportation-related CO2 emissions, but it would require a rapid expansion of charging infrastructure and a significant shift in consumer behavior towards EVs. Improving energy efficiency in buildings and appliances and increasing the use of renewable energy sources could also make a significant impact on CO2 emissions, but again, it would require policy changes, infrastructure investments, and the deployment of new technologies. Planting trees and implementing carbon capture and storage (CCS) could also contribute to reducing CO2 emissions, but these measures alone are unlikely to be sufficient to achieve a 50% reduction by 2030. Overall, reducing CO2 emissions by 50% by 2030 would require a rapid and concerted effort across multiple sectors, including transportation, energy, and industry, and would likely involve significant policy and infrastructure changes, as well as the deployment of new technologies.
From this point the conversation became rather circular, with the system continually rephrasing the answer it had already given in different ways, depending on how I asked the question. More than once it said;
A: However, it is generally understood that the increased deployment of low-carbon technologies, such as EVs, solar PV, and advanced nuclear technologies, can make a significant contribution to reducing CO2 emissions.
A: However, if these technologies were widely deployed and powered by low-carbon energy sources, they could make a significant contribution to reducing CO2 emissions.
But the system was reluctant to offer a numerical assessment even when asked, although eventually it offered a number:
Q: By how much would global emissions fall if EV production, solar PV module production and nuclear power deployment all quintupled by 2030.
A: It is difficult to estimate by exactly how much global carbon dioxide (CO2) emissions would fall if electric vehicle (EV) production, solar PV module production, and nuclear power deployment all quintupled by 2030, as it would depend on a number of factors, including the current level of emissions, the specific technologies being deployed, and the mix of energy sources used to power these technologies.
Q: Can you make a numerical estimate of the likely reduction in emissions?
A: It is difficult to make a numerical estimate of the likely reduction in carbon dioxide (CO2) emissions that would result from quintupling electric vehicle (EV) production, solar PV module production, and nuclear power deployment by 2030, as it would depend on a number of factors, including the current level of emissions, the specific technologies being deployed, and the mix of energy sources used to power these technologies.
Q: Can’t you make an estimate based on some simple assumptions?
A: It is possible to make a rough estimate of the likely reduction in carbon dioxide (CO2) emissions that could result from quintupling electric vehicle (EV) production, solar PV module production, and nuclear power deployment by 2030 based on some simple assumptions. However, it is important to note that such an estimate would be highly speculative and would not take into account a number of important factors that could affect the actual reduction in CO2 emissions. For example, if we assume that:
. . . . and we further assume that these technologies are able to completely displace the higher-emitting technologies they are replacing, we can estimate the potential reduction in CO2 emissions that could result from these measures. Under these assumptions, if EV production, solar PV module production, and nuclear power deployment were all quintupled by 2030, it is possible that CO2 emissions could be reduced by a rough estimate of around 15-20%. However, it is important to note that this is a highly speculative estimate and the actual reduction in CO2 emissions could be higher or lower depending on a range of factors.
Q: Do you mean 15-20% of total global anthropogenic CO2 emissions?
A: Yes, the estimate of a 15-20% reduction in carbon dioxide (CO2) emissions that I provided is based on total global anthropogenic CO2 emissions, which are emissions that are produced by human activities, such as the burning of fossil fuels and deforestation.
There we have it; after some pushing the AI system was able to determine that quintupling (5x) the current deployment of solar PV, EVs and nuclear power generation would give us a 15-20% reduction in emissions by 2030, although this is still along way from the 45% goal of the Glasgow Climate pact. With that in mind I asked a follow-up question.
Q: So what extra steps would need to be taken to get a 45-50% reduction in global emissions?
A: Increasing the use of renewable energy, improving energy efficiency, planting trees, electrifying transportation and making use of carbon capture and storage: Carbon capture and storage (CCS) is a technology that captures CO2 emissions from power plants and other industrial facilities and stores them underground. CCS has the potential to significantly reduce CO2 emissions from these sources, but it is not yet widely deployed.
I was back in the same loop I had started with in that ChatGPT didn’t want to be specific on the measures. After some additional pushing I hadn’t progressed out of the loop, but one extra important insight emerged.
A: Implementing carbon pricing: Establishing a price on carbon, through mechanisms such as a carbon tax or cap-and-trade system, can provide incentives for businesses and individuals to reduce their CO2 emissions.
At least as a set of basic insights and discussion points, ChatGPT seems to have a grip on the climate issue, but it is far from making clear recommendations on what needs to happen, when and at what scale. That still seems to be a job for the energy modelling community . . . . . at least for now!!
ChatGPT is an open system to try and use, which is what I have done. This post isn’t meant as a recommendation or a criticism of ChatGPT, but just a look at how AI is developing.
Happy New Year!
Corporations that connect all their relevant data within a unified AI platform — not just ESG and financial data streams — will benefit the most. When corporations can see the entire picture and understand all the ESG metrics, only then will we produce goals that provide meaningful change and avoid the worst effects of climate change.
At the societal level, AI platforms are also starting to emerge and they are being designed to tackle a broad range of issues and questions by being introduced to huge amounts of data, such as you and I might come across if we sit all day and trawl through hundreds of internet sites and pages. The skill of the AI platform will be to integrate this data and use it to derive clear answers to questions and offer proposals for change so that progress can be made. Ideally the systems should develop insights that are not obvious to humans looking at the same data, perhaps because of the amount of data presented, which the AI system can easily handle.
Late last year one of the newest versions of such a system was made available for public testing. It is called ChatGPT and can be accessed here. To be fair this is a system that is designed more to engage in a conversation than to solve the world’s problems, but it does still draw on an immense data resource in compiling its responses. As such, I thought it would be interesting to engage with the system over the Christmas break and see what conclusions it might draw from a discussion on emissions mitigation. Note that the answers below are abbreviated, simply to limit the length of the blog post.
I started with posing a simple question;
Q: Atmospheric CO2 is now over 420 ppm. Why is this a problem?
A: Rising temperatures, changing climate, ocean acidification and therefore overall negative impacts on human health. Overall, it is important to reduce our greenhouse gas emissions and limit the concentration of CO2 in the atmosphere in order to mitigate the negative impacts of climate change and protect the health of people and ecosystems.
Q: So how might carbon dioxide emissions be reduced very quickly, say halved by 2030?
A: Increasing the use of renewable energy, improving energy efficiency, planting trees, electrifying transportation and making use of carbon capture and storage: Carbon capture and storage (CCS) is a technology that captures CO2 emissions from power plants and other industrial facilities and stores them underground. CCS has the potential to significantly reduce CO2 emissions from these sources, but it is not yet widely deployed.
So ChatGPT certainly understands the issue and is clear on the direction of travel. It also sees the importance of CCS without any prior reference to the technology by me (other than it may have read my blog posts as part of its learning), so that’s a good start. But getting it to be specific on how quickly these technologies could be scaled and how much difference they might make by 2030 turned out to be much more difficult.
Q: How quickly would these steps need to be scaled to cut emissions in half by 2030?
A: The rate at which these measures would need to be scaled up in order to cut carbon dioxide (CO2) emissions in half by 2030 would depend on the current level of emissions and the specific measures being implemented. For example, transitioning to electric vehicles (EVs) and public transportation could make a significant dent in transportation-related CO2 emissions, but it would require a rapid expansion of charging infrastructure and a significant shift in consumer behavior towards EVs. Improving energy efficiency in buildings and appliances and increasing the use of renewable energy sources could also make a significant impact on CO2 emissions, but again, it would require policy changes, infrastructure investments, and the deployment of new technologies. Planting trees and implementing carbon capture and storage (CCS) could also contribute to reducing CO2 emissions, but these measures alone are unlikely to be sufficient to achieve a 50% reduction by 2030. Overall, reducing CO2 emissions by 50% by 2030 would require a rapid and concerted effort across multiple sectors, including transportation, energy, and industry, and would likely involve significant policy and infrastructure changes, as well as the deployment of new technologies.
From this point the conversation became rather circular, with the system continually rephrasing the answer it had already given in different ways, depending on how I asked the question. More than once it said;
A: However, it is generally understood that the increased deployment of low-carbon technologies, such as EVs, solar PV, and advanced nuclear technologies, can make a significant contribution to reducing CO2 emissions.
A: However, if these technologies were widely deployed and powered by low-carbon energy sources, they could make a significant contribution to reducing CO2 emissions.
But the system was reluctant to offer a numerical assessment even when asked, although eventually it offered a number:
Q: By how much would global emissions fall if EV production, solar PV module production and nuclear power deployment all quintupled by 2030.
A: It is difficult to estimate by exactly how much global carbon dioxide (CO2) emissions would fall if electric vehicle (EV) production, solar PV module production, and nuclear power deployment all quintupled by 2030, as it would depend on a number of factors, including the current level of emissions, the specific technologies being deployed, and the mix of energy sources used to power these technologies.
Q: Can you make a numerical estimate of the likely reduction in emissions?
A: It is difficult to make a numerical estimate of the likely reduction in carbon dioxide (CO2) emissions that would result from quintupling electric vehicle (EV) production, solar PV module production, and nuclear power deployment by 2030, as it would depend on a number of factors, including the current level of emissions, the specific technologies being deployed, and the mix of energy sources used to power these technologies.
Q: Can’t you make an estimate based on some simple assumptions?
A: It is possible to make a rough estimate of the likely reduction in carbon dioxide (CO2) emissions that could result from quintupling electric vehicle (EV) production, solar PV module production, and nuclear power deployment by 2030 based on some simple assumptions. However, it is important to note that such an estimate would be highly speculative and would not take into account a number of important factors that could affect the actual reduction in CO2 emissions. For example, if we assume that:
- All EVs are powered by electricity from low-carbon sources and replace gasoline-powered vehicles
- All solar PV modules are used to displace coal-fired power plants
- All nuclear power is used to displace coal-fired power plants
. . . . and we further assume that these technologies are able to completely displace the higher-emitting technologies they are replacing, we can estimate the potential reduction in CO2 emissions that could result from these measures. Under these assumptions, if EV production, solar PV module production, and nuclear power deployment were all quintupled by 2030, it is possible that CO2 emissions could be reduced by a rough estimate of around 15-20%. However, it is important to note that this is a highly speculative estimate and the actual reduction in CO2 emissions could be higher or lower depending on a range of factors.
Q: Do you mean 15-20% of total global anthropogenic CO2 emissions?
A: Yes, the estimate of a 15-20% reduction in carbon dioxide (CO2) emissions that I provided is based on total global anthropogenic CO2 emissions, which are emissions that are produced by human activities, such as the burning of fossil fuels and deforestation.
There we have it; after some pushing the AI system was able to determine that quintupling (5x) the current deployment of solar PV, EVs and nuclear power generation would give us a 15-20% reduction in emissions by 2030, although this is still along way from the 45% goal of the Glasgow Climate pact. With that in mind I asked a follow-up question.
Q: So what extra steps would need to be taken to get a 45-50% reduction in global emissions?
A: Increasing the use of renewable energy, improving energy efficiency, planting trees, electrifying transportation and making use of carbon capture and storage: Carbon capture and storage (CCS) is a technology that captures CO2 emissions from power plants and other industrial facilities and stores them underground. CCS has the potential to significantly reduce CO2 emissions from these sources, but it is not yet widely deployed.
I was back in the same loop I had started with in that ChatGPT didn’t want to be specific on the measures. After some additional pushing I hadn’t progressed out of the loop, but one extra important insight emerged.
A: Implementing carbon pricing: Establishing a price on carbon, through mechanisms such as a carbon tax or cap-and-trade system, can provide incentives for businesses and individuals to reduce their CO2 emissions.
At least as a set of basic insights and discussion points, ChatGPT seems to have a grip on the climate issue, but it is far from making clear recommendations on what needs to happen, when and at what scale. That still seems to be a job for the energy modelling community . . . . . at least for now!!
ChatGPT is an open system to try and use, which is what I have done. This post isn’t meant as a recommendation or a criticism of ChatGPT, but just a look at how AI is developing.
Happy New Year!