Leveraging AI and ChatGPT for Climate Change

Why I am an Optimist

I hear a lot of pessimism nowadays. People hold a deep fear around climate change and the direction AI is developing. I understand that. One must be naïve or wilful not to see that climate change is already creating serious challenges around the world. In the same vein, it is easy to see how AI can either be used for negative purposes, or possibly evolve to pose a threat to humanity.

I don’t share this view.

Yes, climate change is a real challenge. It is obvious to anyone looking at the statistics and the fit between some climate prediction models and the actual evolution of the climate that it can lead to untold misery to millions. The reason for my lack of pessimism, or dare I say, optimism, about the future, is a combination of two factors.

Global population is declining — There is an accelerating decline in global population. Some countries, like Russia, have already began that trend a few years ago. This is not only the story of the affluent West but can be seen even in China where there is a robust debate about the actual population of the country and the pace of its decline. In fact, some predictions I have seen claim that the population of China may halve by 2050.

The AI singularity — This is a term introduced by Ray Kurzweil that describes the inflection point beyond which a general AI, also known as Artificial General Intelligence (AGI), becomes more intelligent than people. It is expected that if that event occurs, the AGI’s intelligence will exponentially grow beyond that point in a flash as it begins training itself. There are two main schools of thought around this event:

• The AGI will become so intelligent and powerful so quickly that it will treat humans the same way we treat ants. That’s not a great outcome. But to be boringly practical, if we become the ants, and the AGI turns into the metaphorical giant that squashes humanity underneath its metaphorical boot, then we will have nothing to worry about. We will be dead. This is not to sound cavalier about such an event, but rather it comes from the belief that humanity does not have the capacity, at this stage, to put in place the checks and balances required to limit the development of AI to avert this scenario.
• The AGI will work together with humans for a better future. That’s where I am putting my money on. And that is why I am optimistic.

A declining population will reduce the pressure on the environment and on humanity’s need to produce food and shelter. A friendly AGI that will be incredibly intelligent will have the capacity to develop the solutions required to both protect humans and restore the planet to a human life-friendly balance.

(BTW, one of the questions I am being asked is whether AI is about to become conscious. Based on my own research I don’t believe that this is something we will see soon. My understanding of consciousness implies that it is bound to specific quantum effects of the biological structure of life and is not information theoretic in nature.)

AI and Climate Change

Humanity is deploying massive resources in response to climate change. In. 2019–2020 the world spent $632 billion on climate mitigation and adaptation (https://makechange.aspiration.com/how-much-money-to-help-climate-change/). Based on 2019 World Bank estimates $90 trillion need to be spent by 2030. This is a lot of money. Any improvement in efficiency around spending these kinds of sums will lead to significant savings. Artificial Intelligence (AI) is emerging the partner that can help fight this battle.

Traditional AI offers data analysis, predictive modelling, and decision support. These can be used on a range of topics from forecasting extreme weather events to optimizing renewable energy usage. Some estimate that AI can save 10–20% of the required expenditure. Given the scale this translates to quite a bundle.

There are two main efforts in the climate space and AI can contribute to both:

Climate Adaptation — Adaptation strategies in climate change reduce vulnerability to the adverse effects of climate change. AI can help understand and predict climate patterns. For instance, it can process vast amounts of data from satellite images and weather stations to forecast droughts and floods, enabling proactive disaster management and resource allocation.

Some other examples include precision farming that uses data to optimize water usage, soil health, and crop yields. Another area is urban planning. AI can help design buildings and cities that can withstand extreme weather events.

Climate Mitigation — Most efforts in the climate space have been focused on reducing or preventing the emissions of greenhouse gases. For example, AI can help in the transition to renewable energy by optimizing the grid integration of solar and wind power and enhancing energy efficiency in various sectors. Another use is improving the monitoring deforestation and carbon emissions.

In the remainder of this article I focus on climate adaptation.

Can ChatGPT be Useful as a Climate Change Adaptation Tool?

ChatGPT has been around for over a year. Large language models (LLMs) are ubiquitous and are in common use nowadays, especially in textual and visual content generation. Its main pitfalls are bias and fantasy. Given these, can it be a useful tool for climate adaptation? To interrogate this I engaged with ChatGPT in a conversation that focused on an area I have little knowledge about: climate change driven food scarcity in Pakistan with an emphasis on wheat. I thought it would be interesting to engage with ChatGPT on a new area for me and see how far we can dive into the issues and provide suggested steps the Pakistani government can take to mitigate the impact of the challenges posed by climate change. I believe that this example demonstrates how ChatGPT can provide policy directions and insights into climate adaptation actions.

The full transcript of the ChatGPT conversation can be found in this link: https://drive.google.com/file/d/1f6iZWpVs15BQhJe7NZlFaJl1QrKfHxC6/view

Case Study: Using ChatGPT to Investigate Pakistan’s Adaptation Policies

I began the discussion with a very general prompt:

I want to know the general predictions for changes in temperatures and precipitation for Pakistan in the next 50 years based on various scenarios. Provide an overview.

ChatGPT provided 6 challenges Pakistan will face due to climate change. I asked it to focus on the impact on agriculture in the next 50 years. The first point it raised was the rise in temperature and its impact on crop yields:

CHAT GPT TEXT — — — — — — — — — — — — — — — — — — — — — — — — — — — — —

Temperature and Crop Yields:

- Rising temperatures can have a direct impact on crop yields. Many crops have optimal temperature ranges for growth, and exceeding these can reduce yields. For instance, heat stress can impair crop development, leading to lower productivity.

- Certain crops may become less viable in traditionally cultivated areas, necessitating shifts in agricultural practices or crop varieties.

—————————— — — — — — — — — — — — — — — — — — — — — — — — — — — — —

I continued by requesting a decade-by-decade breakdown of the impact of climate change on wheat crop yields for 2 and 3 degrees increase scenarios. Before extracting meaningful responses I had to request the system to estimate the parameters needed to generate an answer as ChatGPT did not have the granular climate modelling data required.

Given these parameters ChatGPT estimated that by the 2050s 20–30%, or more, of the population of Pakistan may face wheat related food shortages.

Once we established the possible extent of the problem we now turned to the proposed solutions. I asked ChatGPT the following:

Given these scenarios, and operating from a role of an expert in climate adaptation, agronomy, water management, and plant genetics, provide a detailed list of all the possible adaptation activities that can be taken to in Pakistan to reduce the impact of climate change on wheat crop yields.

It came up with 9 different strategies (The details can be found in the transcript):

CHAT GPT TEXT —————————————————————————————

1. Plant Breeding and Genetics
2. Agricultural Practices
3. Irrigation and Water Management
4. Technology and Innovation
5. Education and Extension Services
6. Policy and Infrastructure
7. Risk Management and Insurance
8. Sustainable Land Use Planning
9. International Cooperation and Knowledge Sharing

———————————————————————————————————————

It seemed quite broad to me. To be able to hone in on more practical approaches I asked what other countries face similar issues and what are the measures they are taking to address them. ChatGPT listed 10 countries with 5 climate adaptation areas in each country. To extract relevant information to Pakistan I now asked it to assess these projects and rank them in order of suitability for Pakistan. This is the prompt I used:

Create a set of criteria to assess the real world impact of the projects described above, present the criteria and the rational for choosing them including a criteria that assess the suitability of these initiatives to Pakistan, and then apply them for each of the projects for the 10 countries described above. Rank the projects for each country, and list them in order of impact and suitability for Pakistan.

After providing this information the system suggested 5 ranked categories of initiatives in other countries that are suitable for Pakistan’s climate adaptation strategy:

CHAT GPT TEXT — — — — — — — — — — — — — — — — — — — — — — — — — — — — —

1. Solar-Powered Irrigation (from India): Addresses water scarcity and energy issues.

2. Heat-Tolerant Crop Varieties (from India): Directly relevant to Pakistan’s increasing temperature issues.

3. Drip Irrigation (from Egypt): Efficient water use is crucial for Pakistan’s agriculture.

4. Agroforestry Practices (from Nigeria): Enhances biodiversity and soil health.

5. Water Conservation Initiatives(from India): Critical for managing Pakistan’s water resources.

CHAT GPT TEXT — — — — — — — — — — — — — — — — — — — — — — — — — — — — —

I felt we were getting somewhere. To make it more practical I asked the system to suggest which are the regions in Pakistan that are most suitable for climate adaptation interventions and which strategies are the best for these regions. ChatGPT identified Punjab, Sindh, and Balochistan provinces as the top three candidates.

In the final step I identified solar powered irrigation and water conservation as the two most important areas for action and asked ChatGPT to issue an implementation plan:

Solar powered irrigation and water conservation appear important in the above list. For these two initiatives provide two plans of action that will assist and administrators of these region to plan and implement these projects, assess the required costs and technology, and identify the top 5 challenges they need to face in implementing these two initiatives

It provided a plan for these two initiatives that had some level of detail. Additionally, ChatGPT identified the top challenges in implementing these strategies, such as the need for significant initial investment, maintenance requirements, and ensuring farmer adoption.

Broader Implications for Policy and Decision-Making

My experience with using ChatGPT in this context convinced me that AI can be a powerful tool in climate change policy formulation and decision-making. I could see how I could have delved deeper into each of the topics and gleaned more knowledge.

I can see how policymakers, environmentalists, and stakeholders, can benefit from using tools like ChatGPT to gain insights and develop informed strategies to combat the impacts of climate change.

Final Words

My experience in using ChatGPT to understand the climate issues for an area I was not familiar with and devising a climate adaptation strategy was positive. I would still check the final outcomes with a specialist, or supporting literature, just to remove the possibility of fictional responses creeping into the dialogue.

This approach can accelerate decision making, planning, and implementation of climate adaptation and mitigation strategies.

ChatGPT is only a small portion of the benefits AI can provide in this field and future articles will examine other approaches.

I can harness ChatGPT and other AI tools to help your company to cut the time it takes to explore new areas and craft new strategies. Contact me at ashimansky@gmail.com to find out more.