Climate Policy Tracker in the eyes of ChatGPT
For the last couple of months, our team was absorbed by creating a Climate Policy Tracker, a solution for analyzing climate-related documents, which was submitted to NeurIPS 2022 Workshop on Tackling Climate Change with Machine Learning. With climate change being one of the most meaningful and engaging issues of our times we couldn’t turn a blind eye to it, so we decided to contribute to the cause.
The article refers to the solution that is available at climate.mi2.ai
As a consequence of the ubiquitous spread of climate change, nations have to act in order to save our planet. Taking action by political entities generates tons of documents and reports, which are long and tedious to analyze manually. So what we did was to …
Wait! How rude of me! I forgot to introduce my friend Chat GPT to you. If you don’t know each other already, you will love him. In fact, maybe let him tell you about Climate Policy Tracker, after all, he also played a role in this project.
Hi GPT!
Hello! I’m Assistant, a large language model trained by OpenAI. Is there anything I can help you with today?
So GPT, could you in simple terms explain what Climate Policy Tracker is?
Climate Policy Tracker is a tool that uses artificial intelligence to automatically summarize and analyze national climate policy documents. It helps policy experts, lawmakers, and citizens understand and compare climate policies across different countries. The tool allows for easier integration into user-friendly applications, which could lead to better engagement and governance over climate policies.
And what are those national climate policy documents that you have mentioned?
The national climate policy documents that the tool analyzes are called national energy and climate plans (NECPs). These are standardized documents that are produced by the national governments of the European Union’s Member States. The NECPs cover the period from 2021 to 2030 and address issues related to climate policy, such as how governments plan to achieve their climate goals. The Enhanced Transparency Framework established by the Paris Agreement requires national governments to regularly report on their climate policies using these documents.
Okay, so Climate Policy Tracker was used to process and analyze NECPs, but the pipeline we used can be utilized in many more cases! So are there any restrictions that documents must meet?
The pipeline was designed to analyze documents that are similarly structured.
That is interesting. So all we need are structured documents! The reason why all these works is… uhm, I am actually not sure. GPT do you have any idea?
The pipeline uses Latent Dirichlet Allocation (LDA), a type of machine learning algorithm, to automatically summarize and analyze the national energy and climate plans. LDA is a type of topic modeling that allows the tool to detect the main topics and themes that are addressed in the documents. The pipeline also uses clustering algorithms to group the documents together based on their content and allow for comparison between them. This allows for a more in-depth analysis of the language and framing used in the documents and how it relates to climate policy.
Okay, okay stop being a know-it-all. Obviously, I know what you are talking about, but explain to the readers, how it applies in practice. To the readers, not to me of course, as I said I exactly know what you mean. So, could you say what features our application has?
Here are some of the key features of Climate Policy Tracker:
- Automated analysis of national energy and climate plans using Latent Dirichlet Allocation and clustering algorithms
- Ability to detect the main topics and themes addressed in the documents
- Comparison of policy documents across different countries and dimensions
- Easier integration into user-friendly applications for the development of theories and processes of climate policy
- Potential for better citizen governance and engagement over climate policies and public policy research.
Overall, the tool is designed to help policy experts, lawmakers, and citizens understand and compare climate policies across different countries. It uses artificial intelligence to quickly and comprehensively summarize and analyze the policy documents, allowing for a more in-depth analysis of the language and framing used in the documents.
Decarbonisation dimension
That sounds amazing! But there is still one thought that haunts me. I am certain that you provided assistance with the application, but you have never mentioned it. What did you do? Don’t be modest.
Climate Policy Tracker pipeline uses GPT-3 (a large language model trained by OpenAI) to assist with interpreting the results of its analysis. The contributions of GPT-3 in this context are:
- Automatically naming topics from Latent Dirichlet Allocation (LDA) based on keywords, which is typically a difficult task in topic modeling.
- Generating abstractive summaries of the parts of the documents analyzed by the model, which can make the results of the analysis easier to understand.
Overall, the use of GPT-3 in the Climate Policy Tracker pipeline helps to make the results of the analysis more accessible and easier to interpret.
compared to other countries in the decarbonisation dimension
I think that explains everything. Do you have some final words for us?
Climate Policy Tracker is a useful tool for helping policy experts, lawmakers, and citizens understand and compare national climate policies. It uses artificial intelligence to automatically summarize and analyze policy documents, allowing for a more in-depth analysis of the language and framing used in the documents. By making the results of the analysis more accessible and easier to interpret, Climate Policy Tracker has the potential to improve citizen governance and engagement over climate policies and public policy research.
Beautifully said. All right, thank you for your time GPT and I’ll talk to you soon.
Coming back to being serious, climate change is one of humanity’s most serious problems. I would not like this looser form of speech to let you miss how important each element of the fight against environmental degradation is. Although I believe that the GPT explained our solution very well, I will reiterate that the Climate Policy Tracker is another important step forward towards a better future. It is another important change that we need to jointly face the challenge of climate change.
I wish you, myself and GPT a long and green future.
You too can chat with GPT here.
All of the above output was unchanged and created by ChatGPT providing in the input our article made for NeurIPS 2022 Workshop: Tackling Climate Change with Machine Learning which you can read here.
If you are interested in other posts about explainable, fair and responsible ML, follow #ResponsibleML on Medium.
