AI in Fundraising 101 – What is it, how is it used, what’s ahead?

Dave Lyndon is Dataro’s CTO and Co-Founder. He has been a software engineer, specializing in databases and machine learning for over 10 years. Dave loves working on state-of-the-art technology for awesome causes and has written this AI in fundraising 101 as an explainer of what AI is and how it can be applied in the context of fundraising. In this blog, we’ll also look at the different types of AI technology, how it works in theory and practice and understand new developments like GPT. The future is closer than you think!

What is Artificial Intelligence (AI)?

Artificial Intelligence (AI) is a field of research and development that has existed since the early days of computers in the 1940s. The goal of AI is to create computer systems that can replicate or exceed aspects of human or animal intelligence. 

Usually when we’re talking about artificial intelligence, we’re not talking about a robot that can see, think, act, and do things exactly like a human (like the Terminator movies). Instead the intent is to use a computer to solve a particular problem or perform a particular task as well as a human can, or better. 

An early challenge in AI was building a program that could play chess. But, these days AI is used in 1000s of different applications. From content recommendation systems (like Netflix), to manufacturing, medical image analysis, spam detection and even in the development of driverless cars, AI is increasingly a part of everyday life.

What are the different types of AI (terminology explained)?

There’s a lot of terminology and jargon people use when talking about AI. So let’s clarify some of the terminology used.

In short, AI is the overarching umbrella term given to the development of computer systems that can perform tasks that would typically require human intelligence, such as understanding natural language, recognising patterns, making decisions and learning from experience. This diagram articulates the different subsets of AI and how they relate to each other. 

Rule-Based Systems

Early work in AI was rooted in ‘rule-based systems’. Essentially, this is where a person analyzes a problem and tries to figure out the rules and patterns before explicitly coding those rules into a computer program. However, it was found that these systems were quite fragile in practice. Real world problems are often messy and there are all kinds of edge cases which are difficult to discover and develop rules for. 

Machine Learning

Machine Learning (ML) differs from rule-based approaches in that instead of a person analysing a problem and developing a solution, an algorithm is used to analyse historical data and automatically discover the underlying rules and patterns.

Machine learning is itself a huge area of active research. Owing to the ongoing increases in the availability of data and computer power, ML-based approaches are increasingly successful in real-world applications. Today, in almost all situations, when someone is discussing AI, they are referencing an ML system.

Deep Learning

Deep Learning (DL) is a sub-field of Machine Learning that makes use of a class of algorithms known as ‘deep neural networks’. Neural networks are learning algorithms that are loosely modelled on the brain in the sense that they have many independent, interacting units or neurons. Deep neural networks use many millions (or billions) of neurons arranged in a deeply layered structure. These kinds of algorithms have risen to prominence since 2012 as important developments in technology and theory have enabled them to be uniquely effective at working with complex, unstructured data such as images, video, sound and human language.

Large Language Models (LLMs i.e. ChatGPT, Bard, etc,…)

Recent work in deep learning focusing on very large scale models trained on enormous databases of textual data have brought forth startling improvements in the quality of models dealing with natural language tasks. These Large Language Models (LLMs) are able to interpret and answer questions, perform translations and summarise information at an extremely high level, in some cases as good as a human specialist.

If you have played with OpenAI’s ChatGPT, you’d be forgiven for thinking you were speaking to a real human and not a chatbot.

Predictive versus generative models

While the underlying algorithms and technology largely overlap, there are broadly two classes of ML systems that a fundraiser may be interested in: Predictive and Generative models.

Predictive models are able to take in large amounts of data and produce accurate predictions about what will happen in future (i.e. given this donor’s history, they are 95% likely to donate in the next 90 days).

Generative models on the other hand, take in a small input (a “prompt”) and produce novel high quality output.

How is AI being used in fundraising?

AI is almost certainly already being used in some of the software or tools that nonprofits are using in their fundraising practices already.

But the biggest opportunity for using AI in fundraising right now is to help you run more efficient fundraising programs deepen the nonprofits’ understanding of their supporter base and how they engage with donors.

At Dataro, we can think about AI technologies as helping fundraisers answer the following questions about your nonprofits’ donors and fundraising activities: 

1. WHO should we contact? 

2. WHEN should we contact them?

3. HOW much should we ask for and HOW should we reach out?

4. WHAT should we say to them?

Machine Learning to answer the “WHO”, “WHEN” and “HOW”

An area where AI is extremely useful is in data analysis and predicting the future behavior of your existing donors. In fundraising, Machine Learning can be used to analyze donor data and identify patterns that can help organizations better target potential donors.

Machine Learning models (like the ones we build at Dataro) are adept at analyzing extremely large datasets – like your donor database – to accurately identify patterns and trends, and predict future outcomes.

In fundraising, this means that Machine Learning models can help nonprofits identify their best donors for various fundraising efforts, based on their likelihood to give a gift if solicited or stewarded effectively. 

Typically and historically, nonprofits perform some form of segmentation to target their fundraising campaigns – that is, taking the database and applying some simple rules to group donors together (like recency, frequency, monetary value). And based on these segments, people would either be included in the audience list for solicitation or excluded. 

The problem with traditional segmentation is that the donor segments used are very broad. While some likely donors will be included in the segment, many donors who are not likely to be receptive to the ask will also be included. This leads to campaign wastage and poor donor experiences. 

Instead of treating donors like a segment, AI helps nonprofits treat donors like individuals. It also helps fundraisers to understand donors better and develop very accurate ideas about what each individual donor is likely to be interested in and what types of campaigns or asks they are most likely to be receptive to. 

But predictive AI models don’t just identify which donors are most likely to give. They can also be used to tell fundraisers:

• who and when to ask for donations
• how many donors to include in your campaign list for best ROI
• how much to ask an individual donor for to improve response rates

Machine Learning can also help organizations optimize their fundraising campaigns. By analyzing data on past campaigns, it can identify which tactics were most effective and which were less so. This allows organizations to adjust their strategies and focus on the tactics that are most likely to result in donations.

Large Language Models to answer the “WHAT”

Up until now AI hasn’t really been able to answer the what in fundraising. That is, what are you going to say to donors (what will you communicate for the best outcomes).

But that has changed with the next generation of generative AI technologies like GPT and other large language models. Using GPT technology and tools like Dataro’s new AI Assist, fundraisers can now generate entirely new pieces of content from scratch, in seconds. 

For example, if a fundraiser has developed a first draft piece of campaign content, like a direct mail letter, they can use large language model tools like GPT and AI Assist to enhance the content or create a multitude of variations for different types of donors across different communications channels. For example, a fundraiser could use these tools to tailor a direct mail letter specifically to lapsed donors, or turn that content into an appeal email targeting young active donors, or even use it to generate content to increase the number of monthly donor conversions via social media.

Developing highly tailored communications can only help improve the donor experience. For the donor, they will be more engaged and have a better relationship with the charity that speaks to them like an individual, not a segment. For the fundraiser, they will be able to create a larger volume of tailored communications in significantly less time, with better fundraising outcomes.

The future of “AI Assistants” in fundraising

Working with hundreds of fundraisers across thousands of fundraising campaigns, we appreciate how busy and under-resourced fundraising and development teams are. Even the most competent fundraisers are crying out for help, but given the tight budgets of nonprofits, they will rarely receive it.

But that all changes with the development of digital “AI Assistants”. 

Right now, Dataro’s newest AI Assist product can save digital fundraisers and marketers countless hours writing and tailoring communications. And we have only just started to scratch the surface with what these tools can do for time-poor fundraisers. 

Imagine what life as a fundraiser would be like if they had not one but many of these AI assistants:

• One assistant to analyze your data and report back on campaign results when they are ready. 
• Another assistant to identify opportunities with new donors and alert you. 
• Another assistant to help generate ideas for a new fundraising campaign. 

This is the future of AI in fundraising – where each fundraiser is supported to work to their full potential and help their organizations operate in a much more efficient manner to advance their missions. And its a future that is closer than you think.

In summary

Like the birth of the internet, AI and now GPT represents a transformative milestone in the field of technology and communication. By leveraging these advancements, nonprofits can enhance their ability to connect with donors, improve resource allocation, optimize their fundraising strategies and ultimately raise more funds to support their cause. 

AI truly is a powerful tool that empowers nonprofits and fundraisers to make a lasting impact and create positive change in the world. 

Interested to learn more about leveraging AI in your fundraising?

Get in touch for a personalised demo of our solutions.