ChatGPT for the nonprofit sector: Potential and Pitfalls

The latest technical innovation that has got the internet abuzz is ChatGPT, so I wanted to write about what it means for nonprofits. Given that it is a chatbot that can answer pretty much any question, I decided to ask ChatGPT to help me write this blog post to demonstrate how it works. As well as being pleasingly meta, this also shows how it works and its limitations through a sort of human-to-AI interview with my thoughts in parentheses…

JW: Suggest a good title for a blog post introducing ChatGPT

ChatGPT: "Unlocking the Potential of ChatGPT: The Future of Human-Computer Interaction"

(Meh)

JW: Try again, and include what it means for nonprofits

ChatGPT: "ChatGPT: Transforming Nonprofit Communications through Advanced Language Generation"

(Sorry, I fell asleep before reading to the end)

JW: Have another go at writing a title for this blog post, and make it less than 50 characters long

ChatGPT: "ChatGPT for Nonprofits: Potential and Pitfalls"

JW: Sold!

Jonathan Waddingham: Describe what ChatGPT is in a short, punchy paragraph that would open a blog post

(you’ve got to start a blog with a punchy paragraph to capture the readers’ attention right?)

ChatGPT: ChatGPT is a cutting-edge language generation model developed by OpenAI that allows for natural and human-like conversation. It's capable of understanding and responding to a wide range of topics and can be integrated into a variety of applications, such as chatbots, virtual assistants, and language translation tools. With its powerful capabilities, ChatGPT has the potential to revolutionise how we interact with technology and pave the way for new forms of human-computer interaction.

(I wouldn’t exactly call that ‘punchy’, so reader please do bear with us)

JW: Write a paragraph listing 5 ways nonprofits could use ChatGPT

ChatGPT:

1. Chatbots: Nonprofits could use Chat GPT to build chatbots that provide information and assistance to donors and volunteers in a natural and conversational way.
2. Virtual Assistants: Virtual assistants could be integrated with Chat GPT to help with tasks such as scheduling appointments, managing contacts, and responding to common questions.
3. Content Creation: Chat GPT can be used to generate high-quality, engaging content for social media, email campaigns, and other communications.
4. Translations: Chat GPT could be used to accurately and efficiently translate communications for multilingual audiences, expanding the reach of the organisation.
5. Fundraising: Chat GPT can help in creating personalised and compelling fundraising campaigns, leveraging on the model ability of understanding context and crafting persuasive messages.

(Of these, point 3 is probably getting the most attention, e.g. see this Twitter thread. I would also add a point 6. around answering random questions quickly in the way you might normally use Google and a point 7. about manipulating data, for example reordering a list or doing a data cleansing task that might normally take a lot of trial and error in Excel. In fairness… I did only ask for 5 things.)

JW: But what are the potential downsides and limitations of using Chat GPT in those ways? (Is this AI self-aware enough to know that life is not all cupcakes and rainbows ¯\_(ツ)_/¯)

ChatGPT: While there are many potential benefits to using ChatGPT in the ways described above, there are also some downsides and limitations to consider. One potential downside is that using a pre-trained language model like ChatGPT can introduce bias into the system, as the model has been trained on a large dataset of text that may contain biases. Additionally, because the model is trained on a large dataset of text, there's a possibility it may generate content that is not appropriate for certain audiences, or that could be offensive.

Another limitation is that the model may be not fully capable of understanding the nuances of human behaviour, culture, and values. Also, ChatGPT is based on a neural network model and requires significant computational resources to run which could be a limitation for organisations with limited resources.

It's important to remember that ChatGPT is not a panacea and should be evaluated and implemented carefully in the context of the nonprofit's goals and values.

(One downside not mentioned is that the answers are not, whilst factually correct, that engaging to read. They’re often a bit repetitive, and don’t always flow. There’s still a place for humans yet…)

But on a serious note, the ethical considerations around bias in results are very important, particularly if your nonprofit works to support marginalised groups. Underrepresented groups are, by their very definition, not going to be well represented in the AI model’s training data set. And because that data set is so large and not completely knowable, biases will undoubtedly remain. Use with caution, as this answer alone brings up enough issues that would need a book to fully unpack and investigate.

JW: Ok, that's good to know. But is ChatGPT and the underlying technology free?

ChatGPT: The underlying technology behind ChatGPT, such as the transformer architecture and pre-training techniques, is publicly available and can be freely accessed by researchers and developers. However, the pre-trained models released by OpenAI, such as ChatGPT, are not free to use. OpenAI provides access to these models through an API, and usage is subject to certain limitations and costs. Organisations or developers interested in using the model will have to sign up and agree to OpenAI's API terms of service, which include usage limits and pricing tiers based on the number of requests made to the API. Additionally, there may be costs associated with integrating the API into an organisation's existing systems and infrastructure.

In summary, the technology behind ChatGPT is available to the public but the pre-trained models are not free to use. You need to pay for API usage with OpenAI.

(I was looking more for a yes/no answer here… For clarity, the use of ChatGPT is free - for now - but if you want to build your own application using the same technology, it is not. Pricing is based on usage and which AI model you use, and not entirely straightforward to calculate. That being said, it is not expensive to access the API and ask questions like you would with ChatGPT, and the cost will only grow if you are building your own technological applications using their API)

JW: All things considered, is ChatGPT going to be a disruptive technology that brings AI into mainstream workflows and practices, and should nonprofits pay attention to it

(let’s summarise this blog post!)

ChatGPT: ChatGPT is considered a disruptive technology because of its natural language processing capabilities and its ability to generate human-like text. It has the potential to change how organisations interact with customers and automate many tasks related to communication and content creation. The ability to understand and respond to a wide range of topics and its ability to understand context can make it a useful tool for nonprofits.

However, as I previously noted, it's important for nonprofits to consider the limitations and downsides of using ChatGPT. Organisations need to have a clear idea of how they want to use the technology and how it will benefit their specific needs. Additionally, because there is a cost associated with using the pre-trained models, organisations need to consider the return on investment for their use of the technology.

In summary, ChatGPT is a powerful technology that could bring AI into mainstream workflows and practices, and nonprofits should be aware of its capabilities and limitations. It has the potential to streamline operations, enhance communications and bring new opportunities to organisations. However, it's important to evaluate its use case carefully, and be aware of the costs associated with its use, as well as its ethical implications.

(This is a pretty good response. Interesting that it uses ‘I’ to describe itself. This anthropomorphism is an interesting element that makes it feel like you’re talking to a person, rather than a computer)

JW: Do you think this would make a good blog post?

ChatGPT: It can be a good blog post, it will depend on how the content is developed, researched, and presented. The title is a good start, it gives a sense of what the post is about, and it's short and catchy. The content of the post should provide a more detailed overview of the capabilities and limitations of ChatGPT, and how it can be used by nonprofits. It should also include case studies or real-world examples of nonprofits that are using or considering using the technology, and the benefits and challenges they've encountered. Additionally, it's important to address the ethical implications of using the technology, such as the potential for bias, and the need for transparency and accountability.

A blog post that provides a comprehensive and nuanced discussion of ChatGPT and its potential use cases within the nonprofit sector would be interesting and informative for readers.

(Well, it might have been a good blog post if you hadn’t written such a boring last two paragraphs. Thanks for nothing, AI.)

(Depict an interview between a human and a computer, in shades of purple, as an abstract image, via DALL-E)

.

Jonathan Waddingham is the MD of Learning at Lightful and has had a career in nonprofit technology that spans 17 years, five of which he has spent at Lightful. He works with our relationship management, design and development teams to deliver our digital skills training programme, BRIDGE, ensuring we support nonprofits to achieve great digital outcomes.

.

Share your thoughts

What do you make of Chat GPT, and what are the best tips you’ve seen for using it so far? Share on our Twitter and LinkedIn and let us know. Bonus points if ChatGPT does the hard work for you…

Subscribe to our newsletter

Did you find this article useful? Sign up for our newsletter and stay up-to-date with blogs, articles, and events in the nonprofit and philanthropy space.