top of page

Can Non-Profits benefit from AI?

Updated: Jun 22

Is AI just another buzzword, or can it genuinely benefit nonprofit organisations (NFPs) today?

Amidst the hype surrounding AI, it's evident that this technology holds significant potential for various industries, including social services. However, before diving headfirst into the realm of AI experimentation, social service organisations should first get their data ‘house in order’.

The AI Hype

You may have encountered the Gartner Hype Cycle, developed by the tech research firm Gartner, which outlines the typical trajectory of transformative technologies: from an initial phase of over-hyped expectations, through a trough of disillusionment, to eventual practical implementation and benefit realisation.

There’s an argument that we are currently on the downward slope after the peak of expectations - note how many software vendors are trying to integrate Chat GPT (or equivalent) into their applications, often with mixed or disappointing results that aren’t producing practical benefit.

Yet there are also reasons for the hype. There are many applications of AI that can now perform at the level or better than humans - AIs are often tested on their ability to do certain word-based tasks like pass the US lawyers bar exam and they are now better than the average human student in doing this. If you’ve played with AI before, you know how surprising it can be - not just as a summariser of information but also as a tool to generate genuinely new ideas or perspectives - it is a surprisingly creative tool. A chart from Time magazine illustrates this progress.

What Machine Learning (ML)?

Machine Learning (ML) is a foundation technology of artificial intelligence (AI) and involves the use of statistical tools to enable computers to develop models without explicit programming by humans. In practical terms, this means computers can analyse data and identify patterns to make predictions or detect anomalies, such as predicting customer churn or detecting fraudulent banking transactions. Some people call this ‘classical AI’ as it’s been around for many years and is already used by many commercial organisations (think about Amazon’s book recommendation engine for example). 

While ML focuses on training computers to develop models, AI’s Large Language Models (LLMs) take ML to the next level by using techniques like neural nets and deep learning on a massive scale to process natural language data. As they generate new content they are also called Generative AI. These models, such as Chat GPT and LLama, are trained on vast amounts of internet text to predict connections between words, with Chat GPT 3.5 boasting an impressive 175 billion parameters.

Benefits Vs Risks

It’s important to remember that there are many issues with AI at a society and system level including potential impacts on jobs, bias, deep fakes and potential effects on inequality. Those are all issues that deserve attention from the social sector. But there are also issues with what the industry euphemistically calls ‘hallucinations’ in models where they make up facts and are inaccurate or just wrong. Better training of models will probably reduce this, but it currently means AI should always have human review and oversight in an organisational context.

The high profile faux pas of Australian academics providing evidence against the big four accounting firms with AI-generated allegations is a cautionary tale.

The other risk is data security. The key to remember is if you use an AI on a website or App (such as Chat GPT), whatever you feed it goes outside your organisation’s systems to that company’s servers. As a result you should never upload private documents and client information to an external AI. That means no use of AI for analysing casenotes as those are highly sensitive documents, or personal client data, or financial data or anything else private or sensitive. This underscores the need for social organisations to develop robust data governance and data maturity programs.

You should never upload private documents and client information to an external AI

It is possible to run AI models within an organisation's systems but that’s a specialist task to set up and run. Eventually large social organisations will train their own AIs on their internal data to allow for more detailed analysis, just like large commercial organisations are doing today.

Practical Uses of AI Today

The short answer to what social organisations should use AI LLMs for is the same as any knowledge industry - to help speed up and improve written word-based work. But this should be done within a clear governance framework that prevents AI from analysing confidential data.

Example Uses of AI Today:

  • Summarising long reports, documents, public documents (government policies)

  • Drawing insights or conclusions from written documents

  • Generating initial lists of ideas, meeting agendas, training programs

  • Providing 'How to's'

  • Drafting first drafts of emails and memos

  • Assisting in generating marketing and donor relations content

  • Summarising meetings

  • Supporting data analysis

​So while there are some limited tasks that AI can help with today, there is more important work that social organisations need to undertake today - and that’s building the wider data maturity of your organisation.

Build Your Data Maturity Before You Tackle AI

AI is an emergent tool, but it needs to be used within a robust governance framework, according to a data strategy, and built on a strong data culture. Most social organisations should start the following steps before they consider the use of AI.

Where to start:

  • Gain acceptance from leadership on the need to become a data-driven organisation

  • Audit data maturity and systems usage

  • Recruit a cross-organisation data champions group

  • Develop a roadmap for accelerating data use

  • Establish a Monitoring, Evaluation and Learning framework

  • Develop governance, protocols and data roles

  • Enhance data quality

  • Build practical data products for decision making

  • Foster a data-driven culture

  • Start sophisticated data analytics projects, including Machine Learning projects and AI integration, over time

By following these steps, organisations can lay a strong foundation for leveraging AI in the future.

15 views0 comments


bottom of page