AI for Purpose: A practical guide for charities and NFPs

Insights from our Sold-Out Community Event on Ethical and Effective AI Adoption for Charities and NFPs

Artificial Intelligence (AI) is rapidly entering the charity and not-for-profit sector, offering powerful tools to help organisations increase their impact with limited resources. While AI presents significant opportunities, it also introduces new risks. 

AI can reduce days of work to just moments. It can assist with drafting funding applications, analysing data, or responding to supporter queries. In a sector where time and resources are often stretched, tools that enable more to be done with less are understandably appealing. 

However, this creates a tension. Charities are trusted with sensitive data, vulnerable communities, and public or donor funding. As such, they must manage risk carefully. Some organisations have already begun experimenting with AI, sometimes without fully understanding the implications. 

This blog provides guidance on how charities can explore AI safely and effectively. It outlines the current landscape, common use cases, potential risks, and practical strategies for responsible adoption, including key takeaways from Inde’s sold-out community event in Christchurch, where local charities and not-for-profits gathered to discuss how AI can be harnessed ethically and effectively to amplify their missions. 

What is Artificial Intelligence (AI)?

AI is a broad term that can mean different things depending on the context. In most current applications, AI refers to tools powered by large language models (LLMs), which are trained on vast amounts of text to generate human-like responses. 

These tools are already being used to automate administrative tasks, support communications, and improve productivity. In the future, more advanced AI may support decision-making, predict service demand, or help design new programmes. 

Understanding the capabilities and limitations of AI is the first step towards using it responsibly. It is important to distinguish between what AI can do today and what remains speculative or experimental. 

Limitations of AI and Large Language Models 

While AI tools can be incredibly useful, they are not without serious limitations. Understanding these limitations is essential for using them safely and effectively in your organisation. 

• Hallucinations and factual errors: AI systems can generate responses that sound convincing but are factually incorrect. These are known as "hallucinations" and are a well-documented issue. 

• Bias: AI models can reflect and amplify biases present in their training data. This can lead to outputs that are unintentionally discriminatory or misleading. 

• Lack of reasoning: LLMs do not understand logic or causality in the way humans do. They are trained to predict the next word in a sentence based on patterns in data, not to reason or think critically. 

• No understanding of truth: These models do not have a concept of what is true or false. They cannot verify facts or understand the consequences of their outputs. 

• Limited creativity: AI creativity is based on recombining existing patterns. It does not invent new ideas in the way a human might. 

• No real-world awareness: AI has no emotions, values, or situational awareness. It cannot understand context beyond what is provided in the prompt or its training data. 
• Limited user context: AI tools often struggle to maintain context across longer conversations or understand nuanced user needs unless explicitly guided. 

How AI Works 

Current AI reasoning is achieved through techniques such as: 

• Query reformation and decomposition: Breaking down complex questions into simpler parts. 

• Cross-domain pattern recognition: Identifying similarities across different topics or datasets. 
• Context integration: Drawing on multiple sources of information to generate a response. 

This is why AI tools are not designed to say “I don’t know” when they lack information. Instead, they will often provide an answer – confidently - even if it is incorrect. 

Understanding these limitations helps organisations use AI more effectively and avoid over-reliance. It also reinforces the importance of human oversight, especially in sensitive or high-stakes contexts. 

Risks to Consider 

AI adoption introduces both technical and ethical risks. These can be grouped into two categories: 

Hard risks: 

• Data security: Inputting sensitive information into public AI tools can lead to data breaches or violations of data protection regulations. 
• Compliance: Misuse of AI could result in non-compliance with legal or funder requirements. 

Soft risks: 

• Over-reliance: Excessive dependence on AI may reduce critical thinking and human oversight. 
• Bias and misinformation: AI systems can reflect and amplify existing biases, potentially leading to harmful or exclusionary outcomes. 

For charities, where trust and accountability are paramount, these risks must be carefully managed. 

Understanding What You Want to Achieve With AI 

Before adopting AI, it is essential for charities to define their goals. AI should support the organisation’s mission and values, not distract from them. By identifying specific use cases, charities can ensure that AI investments are aligned with their strategic priorities. 

Strategies for Safe and Effective AI Adoption 

Even without access to enterprise licences or sector-specific tools, charities can begin exploring AI in a responsible way. The following strategies can help: 

• Create test-and-play environments: Encourage staff and volunteers to explore AI tools in a safe, structured setting. This builds confidence and helps identify valuable applications. 

• Protect sensitive data: Educate teams on what constitutes confidential information and ensure it is not entered into AI tools. Use DLP (data loss prevention) policies in conjunction with manual labels and classifications to ensure no sensitive data is used.  

• Develop a comprehensive AI policy: Establish a clear organisation-wide policy on which tools can be used, by whom, and for what purposes. By detailing requirements such as training and use case scenarios it will give employees and volunteers a framework to experiment with while keeping protective measures intact and avoiding accidental misuse.  

• Start small: Pilot low-cost or free tools before committing to larger investments. This allows for learning and adjustment without significant risk. 

• Develop internal champions: Identify early adopters who can support others, share insights, and promote best practice across the organisation. 
• Train your team: As with every new technology there will be early adopters and laggards. Encourage practical trainings on AI around ethics, use cases and data protection. 

These steps can help charities build a culture of safe experimentation, where innovation is encouraged but not at the expense of trust or integrity. 

Conclusion 

AI is not just for large corporations or tech companies. It has real potential to help charities and not-for-profits work more efficiently, reach more people, and increase their impact. 

However, with that potential comes responsibility. By understanding what AI is, setting clear goals, assessing risks, and implementing practical safeguards, charities can explore AI in a way that is ethical, effective, and aligned with their mission. 

The future of AI in the charity sector is about thoughtful, informed adoption. With the right approach, AI can become a valuable tool in advancing social good. If you’d like advice on how to leverage the benefits of AI in your organisation - without the risk - please contact us, we’re here to help.