Four ways local organisations can use AI to support locally led adaptation and build resilience

As the sudden growth of artificial intelligence tools captures the public imagination, Sam Greene gets ChatGPT’s help to explore how AI can make a difference to local communities’ climate work.

Sam Greene's picture
Insight by 
Sam Greene
Sam Greene is a senior researcher in IIED's Climate Change research group
25 May 2023
View of the Earth from the space

While artificial intelligence can be used to monitor and model climate change, the sudden advance in AI large language models also provides practical opportunities for local organisations working to adapt to climate change (Photo: NASA via Unsplash)

The explosion of accessible artificial intelligence (AI) tools such as ChatGPT or Google’s Bard into everyday life presents new opportunities and challenges across every sector and industry. But what does it mean for those on the frontline of climate adaptation work?  

Beyond ChatGPT– a dialogue-based information tool – AI technology is being adapted for specific purposes such as translation; writing and editing support; image development; website design; data and spreadsheet analysis; and coding advice.  

These tools are cheap to use, accessible to anyone with an internet connection and improving at an incredible pace. So how can they be mobilised to support locally-led climate action? 

For years, organisations in the Majority World, including governments, community business organisations, and small and medium enterprises, have highlighted the problems of complex donor funding conditionalities, language barriers and challenges in managing and analysing data without time and money spent in formal education. Such characteristics reflect neo-colonial tendencies within the aid and climate finance sector.  

Their ability to put locally led adaptation into practice by applying their own expertise to support communities and respond to local priorities is undermined by these frustrating barriers.  

So here are four ways that AI tools can support local NGOs and partners to effectively access finance and influence development discourse.  

(For full disclosure, this article has itself been developed in dialogue with ChatGPT 4.0. Perhaps unsurprisingly, ChatGPT was quite ambitious about its own capabilities (more on that below), so I have thoroughly edited its results for relevance.  

1. Writing proposals and reports  

Reading, understanding and responding to donor calls for proposals, usually in English, is challenging for organisations with great local understanding but with staff without a formal education. Proposal calls are often written in obscure technical language and accompanied by complex addendums explaining financial procedures. 

Turning transformative ideas into formal proposals is the next problem. Writing in the languages and conceptual structures of donors is time consuming and taxing. Local organisations can end up losing out to elites with fluent English but with less connection to community priorities. Time consuming reporting requirements are another longstanding complaint – with government officials or NGO officers filling in forms that many donors never read. 

ChatGPT, using its understanding of context, can turn ideas into suggested text for proposals, which can then be edited and revised with the local context in mind, rather than written from scratch. 

Other tools such as Quillbot can address these challenges, instantly rephrasing chunks of text in a range of different styles. 

2. Building capabilities of community-based and local organisations 

AI can help community-based and local organisations develop their capabilities and understanding of a wide range of topics, including fiduciary management, resilience, monitoring and evaluation, and adaptation options.  

AI tools can highlight ways to reduce fiduciary risk and point out avenues for further exploration – all in conversation with the user. It can explain financial terms in simpler language, and in time provide cheap and high quality translation.  

In conversation on how to improve fiduciary management, ChatGPT suggested I establish a chart of accounts, helped me to choose affordable software and identified financial controls I could use. It even translated all that into Swahili, for good measure. 

3. Improving data collection and analysis 

High-quality data is essential for local decision-making in climate adaptation and resilience building. Networks such as Slum and Shack Dwellers International have led groundbreaking work to turn citizen science into actionable data for municipal governments through the ‘Know Your City’ project.  

The ability of community representatives to advise governments with their own data has given them power to shape government activities and act as advisors.  

AI can enhance this process by making analysis more accessible for local organisations that engage communities and collect data. Tools will soon be able to organise raw information and provide analysis based on verbal requests from the user, while also identifying other available datasets and explaining how to integrate them into the analysis.  

Local actors can merge this information with local insights and their understanding of the holistic and interconnected nature of local environments to develop context-specific and community-identified adaptation priorities and plans.  

Further tools, such as Midjourney, can create powerful visuals and infographics that inform local or national public discussion, while AI can also be used to provide creative suggestions for successful marketing – building networks of more effective advocates for community priorities and knowledge.  

4. Simplifying complex policy, financial and legislative documents 

AI can play a vital role in breaking down complex policy, financial and legislative documents, making them more accessible to community members and their representatives. This can contribute to greater transparency and accountability, enabling communities to better understand their rights and provide greater access to resources.  

AI-driven natural language processing tools can democratise technical language and present information in clearer formats and in a range of different languages – providing summaries, enabling communities to search for common criticisms or support for different policies and then to form their own judgements. 

By doing so, AI can ensure that local communities can actively participate in decision-making processes and have their voices heard. 

Addressing risks and challenges in AI adoption 

However, there are risks. And these are significant and many. 

What impact will AI have on the digital divide? Almost 40% of the world’s population did not have a reliable internet connection in 2021, and AI threatens to deepen this divide, increasing the opportunities for the haves and the obstacles for the have-nots.  

Supporting partner organisations to get connected quickly, and avoid being left out, is a fundamental concern.  

We cannot ignore, either, that AI models are not inherently truthful. Information is presented with the authority of great computational power – but these are basically predictive programmes, prone to errors, known as ‘hallucinations’, and don’t have the real world experience to back up its outputs. 

Advice from AI needs to be reviewed and assessed for its legitimacy in context. It can be excellent for editing and idea creation, but that does not make up for personal knowledge, trusted relationships and life experience.  

There’s also a risk that AI language models perpetuate existing power structures and marginalise certain groups. Trained using large amounts of text and images from across the internet, they merely echo the power structures and biases already present in that information, itself a reflection of the biases in society. They are not built to actively combat any form of marginalisation or discrimination such as racism, classism or sexism. .  

In the dialogue I had with Chat-GPT, it argued that “AI developers must prioritise fairness and inclusivity” but this is unlikely without regulation. AI developers, private companies typically based in the global North, have few serious incentives to prioritise the perspectives of local and marginalised voices.  

A critical lens is therefore needed when reviewing anything beyond the simpler tasks such as editing or running calculations. 

AI offers a jump start 

The high impact, low-cost potential of these tools should be catching the attention of any donor or organisation that claims to support locally led climate action.  

With caution, these tools can be an asset to all kinds of local organisations, helping them hurdle some of the most fundamental and frustrating barriers – such as language – so they get to the real business of applying their local knowledge to support communities to deliver locally led adaptation.