There is a disconnect between the richness of the human stories we witness and the poverty of the data we report, which doesn't tell the full story. Our CSOs measure "things" instead of measuring "changes". That is because measuring changes in people's lives takes time and resources.
In this area, Artificial Intelligence (AI) has become an ally. Not to go faster (which it does too), but to help you listen and better understand the people you support and tell their stories. The power of Large Language Models (LLMs) lies in their ability to process data in ways that were simply impossible for most.
The efficiency trap vs. the comprehension revolution
The prevailing narrative about AI is its efficiency. When we hear about tools such as ChatGPT or Gemini, we tend to think of them as ‘text or image generators’ that can help with writing your newsletter in 5 minutes, summarizing your emails, and creating drafts in an instant.
While very useful, this view of AI is limited. Because it's more than a robot that writes to do 'the usual' faster. For CSOs, the paradigm shift of AI is in doing tasks that were 'impossible' or unattainable. As well as having the power to 'see and hear' our world with new senses.
AI can dive in and find patterns in seas of data, transforming dispersed and disconnected information into meaningful data. With the right questions, you can find out what people are worried about. It can support you in developing actions that address those concerns and then measure their real impact.
Using AI to understand stories
Starting to integrate AI into your organization and achieving practices that improve the comprehensive execution of your actions does not require large investments. Here are some examples of how a small NGO can start changing its paradigm with the help of an AI chat.
Enabling inclusion from the start
Situation: You're hosting in-person workshops and need to know if people will need support to make it easier for them to participate. Analyzing open surveys is cumbersome. Using multiple-choice questions with all the possibilities is not practical or accessible. And you can skew the answers according to the interpretation you make of them.
Use AI: If you are planning to organize an event or workshop, and you want to gather any accessibility requirements/special needs from attendees thanks to the Natural Language Processing (NLP), you can analyze open-ended questions such as, Do you need any extra support to make it easier for you to enjoy and learn from the workshop?. By collecting all the answers in a spreadsheet (always without personal data, as discussed further down in the article) ask your AI to:
"Act as an Event Logistics Coordinator. I will provide you with a list of answers from an enrollment form. Your task is to analyze the special needs of the attendees. Please follow these steps:
Group comments by categories.
Identify the specific barriers mentioned.
Indicate how many people mention each barrier.
Propose a viable and concrete logistical solution for each identified barrier."
The AI will process the information exposed in natural language. Revealing needs that were not on our radar. And it doesn't end here...
Situation: At the end of the workshops, you conduct a satisfaction survey using paper and online forms. In addition, you receive emails with answers without the formatting of the questions you designed. You must process all these answers to unify them, as well as analyze the open-ended questions, which complicates it further.
Use AI: You collect responses online, scan paper surveys and group emails into PDFs. With all the data, you attach it to your AI chat and ask it to:
"Act as a Quality and User Experience Analyst. I attach three data sources (paper surveys, Excel surveys, and unstructured emails). Process all the information as a single set of data and generate a report with:
Global Sentiment Analysis: classifies each review as Positive, Neutral, or Negative, as well as generate a bar graph with the total percentages of each one.
Identify and list the 5 most and least valued aspects. Indicating how many people mentioned it.
Group suggestions for improvement by theme."
At the end, it will present the report in the format you requested. You can continue to consult the chat to detect new information or make correlations.
From case studies to storytelling with data
Situation: In the fieldwork, notes are taken in notebooks or mobile phones, photographs and recordings to document an event, or social action. That data can be put aside because it cannot be processed in time and written in a format that shows the complete story.
Use AI: Gather all the information generated during the intervention. Identify each piece clearly and logically. Then, you can proceed to ask your AI to help you structure it:
"Act as a Social Work expert with skills in Social Impact Storytelling. You have a series of raw and fragmented data about 'Juan', a participant in our program. Your goal is to write a Case Study Chronicle that is strictly faithful to the facts. With this narrative structure:
Starting Point: Describe John's initial situation using the data to 'paint the scene'.
The Challenge: narrate the obstacles he faced, based on the notes on his difficulties.
Towards the Future: with a strategic tone, propose a 3-step plan for their employability and labor insertion.
Golden rule: don't invent emotions. If you describe any, it should be based on the data collected."
This approach helps you tell stories with data. Highlighting the human aspect of the intervention, without altering facts.
Simplifying the complex without losing depth
Situation: Every year, we access dozens of reports which can be useful for the objectives of our NGO. These reports can be extensive and technical, making them difficult to analyze. Although they are vital to understand our context, adjust future actions, as well as detect risks and opportunities.
Use AI: Ask your AI to act as an analyst and extract the information from those reports. Directing attention to important content with a format according to your knowledge, or as a tutor who explains it step by step. Try the following with your AI:
"Act as if you were a Strategy and Talent Development Consultant. I attach two reports: one about youth unemployment in Europe (Report A) and another on the impact of AI and future skills (Report B). Your goal is not to summarize them separately, but to synthesize them to find strategic opportunities. Generate a report with a professional but didactic tone, including:
A Cross-Diagnosis: identifying where the two issues overlap.
A Tension Matrix: that contrasts threats, opportunities, and gaps.
Proposals for Action: suggest 3 lines of action for my organization based on the insights detected to support young people in these times of change.
Focus on how we can act as a bridge to close the detected gap"
Leverage these reports to turn them into insights for any organization's mission and help plan for your next strategics actions.
Using AI like the big guys
You don't need a software engineering and data science team to start using AI. Try incorporating AI by prompting into your day-to-day working methodologies. As well as acquiring basic knowledge, you can apply in practice.
Centralize your entity's data. Digitize your documents, AIs understand information contained in images better thanks to their Large Multimodal Model (MML) features. For example, try scanning old documentation.
Educate yourself in Prompt Engineering. Learn how to talk to AI naturally. Know how to ask AI, and do ask AI.
Always verify. Preserve the so-called "human-in-the-loop". AI is not foolproof. It risks "hallucinating", misinterpreting human ambiguity, or cultural nuances. Maintain a constant audit that verifies the results generated by AI.
AI and Data Ethics
Privacy is sacred, and you must make sure to protect your organization’s private information. AIs can learn from what we write to them, and analyzing life stories with AI introduces considerable challenges for a CSO.
Always read the privacy policy. Learn about the type of privacy compliance in the LLM you use.
Do not use personally identifiable data of your beneficiaries in business models. Act with maximum attention towards highly sensitive data.
Anonymize your information before processing it. Remove any data that can identify a specific person.
Pay attention to "Algorithmic Bias". LLMs may adopt biases during and after their training.
Before you rely on an AI tool for big decisions, give it a try. And compare results with other AI tools.
If you need to work with private data, use "Local AI" solutions with your own servers. For instance, with Ollama you can run language models locally to generate text, summarize documents, compose emails, and translate content securely.
The conclusion of the paradigm
We were stuck behind the numbers because listening to hundreds of people was impossible. Now, technology removes that barrier. Most importantly, it helps us put people at the center of all our actions.
Thanks to AI, CSOs can focus on being the main actors of change. Thus, reaching efficiency levels that were not possible before, due to the time and resources constraints. By eliminating these barriers, CSOs can take their impact to the next level.
_______________________________________________________________________________________________________________________________________
Disclaimers
This piece of resources has been created as part of the AI for Social Change project within TechSoup's Digital Activism Program, with support from Google.org.
AI tools are evolving rapidly, and while we do our best to ensure the validity of the content we provide, sometimes some elements may no longer be up to date. If you notice that a piece of information is outdated, please let us know at content@techsoup.org.
This content was created with AI assistance and has been reviewed and edited by Gustavo A. Díaz González.
"Beyond the efficiency paradigm: Using AI to understand our real impact with some practical templates", by Gustavo A. Díaz González, 2025, for Hive Mind is licensed under CC BY 4.0.