From accessible to high end implementations, this is what you can do to improve your impact
In the rapidly evolving field of Artificial Intelligence (AI), charities are finding innovative ways to leverage technology for greater impact. Here James Gadsby Peet, our Director of Digital & Strategy, delves into the diverse applications of AI in charity work, highlighting real-world examples and practical guidance from conversations with our clients and others in the sector.
[Written January 2024 - all the information in this article is liable to change rapidly]
What can be done for cheap or free?
High end AI and Machine Learning projects are not available to most charities given the costs involved. This is because they rely on highly structured data and can only happen using specialist partners and tools, which ensure the highest levels of data security.
But there are all sorts of ways that tools like ChatGPT can help people in charities as long as you bear in mind that the outputs you get are only as good as what you put in.
With any of these tools, you need to:
Train it on your tone of voice
Specify specific biases you’re looking to avoid
Start with good information created by you and your team
Consider the public perception of your work
1. Translating “Business speak”
Many people struggle with the inaccessible language that is inherent in workplaces. ChatGPT can help to translate words into simpler formats for people to understand.
“I use it all the time because some colleagues speak a management language here that my brain struggles to translate. So I take long verbose paragraphs and ask GPT to ‘explain this to an idiot’ or ‘say this without any management speak’.
It can also help them to create their own content that others find easier to comprehend.
At the same time, there often feels an expectation for me to reciprocate with equal verbosity so GPT helps me put my normal speak into business bullshit. As someone who ticks pretty much every box of ADHD, it basically made my working life here SO much easier!”
2. Rewriting content in a different tone
ChatGPT is great at taking a pre-existing piece of content and applying a different tone. You can even get it to research a particular organisation and create something that uses the kind of style that people from that organisation do.
Be sure to:
Train the tool by getting it to read previous content you’re looking to build on
Give it bullet points or original content to work with
Specify the length of what you need it to create
3. Grant Applications
Creating quality grant applications can be hugely time consuming for people, especially if applying for funding is only a small part of a much bigger role. Each funder wants slightly different answers whilst looking for specific pieces of information from you.
ChatGPT can supercharge this process if you give it the right prompts - helping to bring in more money in less time. Anne Bickerstaffe, CEO at CARE Ltd, uses prompts like this to help her:
“Hello ChatGPT, please can you help to answer these funding questions with the correct word count. Use the information that I have given you about our project, and align our project with the funder’s guidelines. Please write in direct, straight forward language that is not flowery. Use as many direct quotes as possible, and specific facts and information about the project. Good answers will have detail about the project which will help the funder to understand the project and approve it for funding. I will now copy the first 4 questions. Thank you”.
Anne recently completed a complex funding application with 17 questions in 13 minutes using ChatGPT. “The answers were so good and so clear that it only took less than 2 hours to edit the final application ready for submitting. As a busy CEO of a small charity, this kind of time saving is incredible.”
4. Interview questions and tasks
Once you’ve spent time crafting a high quality, inclusive job description, ChatGPT can give you a list of ideas for questions and tasks that can be used in an interview to explore someone’s fit for the role.
Use prompts such as:
“ChatGPT, read this job description and say when you’re ready for the next prompt”
“Suggest 6 different tasks that can be used to explore someone’s suitability for this role. Pay particular attention to tasks which do not create barriers for people”
Repeat the process for interview questions.
5. SMART objectives
Once someone has started in a role, setting objectives that help them to understand what success looks like is a crucial step in their development. ChatGPT can give you a long list of ideas, based on a job description and organisational strategy.
6. Debugging code
Developers are finding this kind of prompt a handy shortcut for bug fixing in all different types of code - it’s like having a very patient and always-available teacher.
“Why doesn’t this [insert language] code work”
Paste the code into ChatGPT and you’ll often get back a well referenced and technically accurate result. If you don’t understand what’s being suggested you can ask over and over again for the tool to explain.
7. JustGiving Story Enhancer
JustGiving have invested heavily in creating a new tool for people building fundraising pages to improve the words they use to tell their story.
It puts the user in control of the experience and builds confidence in sharing their unique, personal story that drives fundraising.
Find out more about the tool here - https://blog.justgiving.com/an…
Never share personally identifiable data with ChatGPT
It is extremely important not to share any personally identifiable information with free to use tools such as ChatGPT. High end uses of AI rely on proprietary systems where data security can be absolutely guaranteed, which is not the case for publicly available ones.
When you ask ChatGPT whether you should upload data such as medical information or previous donor behaviours it is clear that this would be a bad idea as it potentially breaches an individual’s right to privacy. If you are using AI tools for spotting patterns or analysing such data, you must ensure it is completely anonymised.
High end AI implementations in charities
There are dozens of high end AI tools being developed - especially in the health sector. Prostate Cancer UK and Moorfields are just two who are using the latest technologies to spot patterns in data that might take humans years to uncover.
There are two widely known examples of AI being used inside charity operations right now - Prostate Cancer UK’s Christmas appeal and Parkinson’s UK social listening project.
Prostate Cancer UK spent lots of time working on their donor data, to try and uncover the best people to send out their Christmas campaigns to. Three datasets that were used covered over 1.5 million customers, over 5.5 million gift transactions, and over 15 million appeal activities. The resulting selections generated over double the ROI of the standard appeal.
During the pandemic, Parkinson’s UK used AI tools to monitor the topics that were most concerning their community. Given the pace at which things were changing, it allowed them to respond quicker and more effectively than ever before. By using a technique called Comparative Linguistics, they were able to identify how the conversations on the forum, helpline and social media groups were changing on a week-by-week basis.
Public Attitudes to AI
The Attitudes to AI survey is a nationally representative survey of public attitudes to artificial intelligence (AI) in Britain by the Alan Turing and Ada Lovelace Institutes. 4,000 people were surveyed and the report shows a nuanced picture. Most people see the benefits but are also worried about AI overriding human judgement in specific situations.
Almost everyone wants some form of regulation - but there is a lot of disagreement about who does that. 43% of 18–24 year olds want companies to self regulate whereas only 17% of 55+ support this.
For people considering the use of AI in their charities, there is very little regulation or guidance on how to deploy it safely. Some of the tools are improving their own self regulation to stop you doing dangerous stuff, but mostly it’s down to you.
A good rule of thumb is asking, could I justify this to a supporter? Rather than trying to find a black and white legal definition as there just aren’t many available.
Setting up an AI community of practice
This article came from a fantastic session that our client Robert Dufton and his team ran at Moorfields Eye Charity - exploring how different parts of the organisation could use AI tools. It served as an excellent way to give people permission to test and learn with these tools.
To so many of us it feels like cheating to be using these new approaches to old challenges. However the organisations that grow their impact and increase their fundraising will be the ones that embrace these new ways of working.
However, there is much to still work out for everyone within each organisation - so our recommendation would be to set up a community of practice about AI. Communities of practice can unite people across an organisation and help them learn from one another whilst pushing forward everyone’s capability.
Find out more about how to set up a community of practice here:
Or get in touch if you’d like to discuss some support in making this happen: email@example.com