Some in the sector think that AI will take over our world, with huge job losses, blah. Other charities don't think it'll impact them at all, because they're brilliant and it's not what they do. We think both are completely wrong.

We think there will be fundamental change, in the same way there was with the advent of the Internet, personal computers and Windows. Taking a lesson from this history, we think the immediate changes are probably over-hyped and the long-term change under-estimated, or at best, as yet unknown.

We think this change will happen quickly and be a bumpy ride, and a major risk for some kinds of charities, if they ignore it. But we also think it's a huge opportunity for a badly under resourced sector, it won't lead to major job losses and the major challenge for us won't be the AI itself, but in making sure we enable access to its potential for everyone, not least the small and marginalised groups, who are far too often left behind.

Generative artificial intelligence (AI) describes algorithms that can be used to create new content, including audio, code, images, text, simulations, and videos.   ChatGPT is the best known, but only one of the new kinds of generative AI systems. 

It's different to existing AI (described above) - it's the AI that creates the content itself using often huge resources taken from the Internet. It is then trained by its controllers on what is and is not a good response. This makes it hugely powerful, but also brings with it not insignificant problems, which is why we don't use it for answering questions and when we do use it, we design in controls to minimise the problems.

ChatGPT only launched in Nov 22 and now has more than 100 million users worldwide.  It's a bit like auto complete on your phone, but it can 'auto complete' and pass university level legal and medical exams.  The tech companies are pouring £s billions into its development and are already beginning to integrate it into browsers and even Excel.  Very soon, pretty much all non-profits will be using it and, even if not, the very widespread use of AI in systems that impact on their work will.  

Whether we like it or not, AI will fundamentally change the sector and in the near term.  

But there are also problems. The new AI systems aren't sentient and can't discern in the way a human can - that's why they struggle with the image above. We can train them to deal with that and to appear very human, but they are not.

They are also 'black box' - we can see what goes in and comes out, but we're not sure what they actually do.  We know they can hallucinate, lie convincingly and make mistakes, and Chat GPT doesn't know about anything that has happened after 2021.  In creating generative AI you must understand its limitations and test it relentlessly, and its outputs must be checked for accuracy, bias and fairness.  For our own AI services, we've designed this in and we're now creating a Charity AI Governance & Ethics Framework, so others can too, but in the rush to make potentially huge amounts of money, how many won't? 

There's also a huge job of work to be done in regulatory terms, because the tech is now far ahead of both the legal and ethics issues it has created. Making legal changes can take a very long time and I'd argue that generative AI isn't inherently unethical, but it reflects our society and its problems. The emerging AI systems have guardrails, but you can still talk your way round these and changing how AI collects and uses data can't fully fix these issues, because it's at least partly a societal, not an AI problem. Fixing these issues will take years and we're likely to be in for a bumpy ride in the meantime.

Ultimately, some aspects may not be fixable. For example, AI struggles with differentiating black faces, because it was primarily trained using white faces and there are other racial biases within the AI data and how it's been created. It may be that we have to accept this for low risk facial recognition, such as unlocking your phone (because you also have a PIN) and accept that it shouldn't be used in policing. The use of AI in criminal justice alone, is likely to result in significant, very important, and probably acrimonious debate for society. We're taking a risk based approach.  If we can't fix something, or we can't be confident that we have and it's important, we won't use it. 

The issues outlined above will impact on society as a whole, but what about sector specific issues? 

There may be questions about the use of AI to create content, such as funding bids.  Our view is that assessing a bid should be about facts and information - the unmet need, the project, its planning and budgets etc, not how well it's written.  Currently, the smaller and marginalised groups lose out to professional fundraisers who are better able to write compelling bids.  The free AI funding bid writer we've created enables those using it to think through what to include in a simple, structured way, but it only uses the information it has been provided with and its ability to draft these well, makes writing funding bids accessible to those who otherwise may not be able to.  Unpaid volunteers with very little time, those who don't know what should be included in a bid or how to structure it and, not least, those for whom English is a 2nd language, or people with learning disabilities or people who just don't have expertise in writing well crafted prose.  It helps to level a currently very unequal playing field and also supports grant makers by helping to ensure that the bids submitted are complete and contain the information they need and want. 

Those deploying AI content will also need to think about legal and Intellectual Property (IP) issues.  In our case, our Ask Me Anything service extensively accesses Crown Copyright material, but this is available to use free of charge under the Open Government Licence.  We also promote the work of 1000s of charities and other organisations that support the sector, often using text from their websites on a 'fair use' basis and actively welcome them submitting changes.  However, you will still have to check that anything you create does not breach copyright and much of what's on generative AI systems, such as imagery software like DALL E, may potentially do so. 

There's been quite a bit of talk about AI taking people's jobs, but we don't think so.  The same was said of the Internet, personal computers and Windows.  There has been huge change, but whilst some jobs were lost, many more we thought would go, simply changed (often for the better), and new jobs were created.  We think generative AI will have a similar impact.   

For example, in the private sector, content writing roles are going already, but some charities will be able to bring agency work back in-house and new roles are already appearing, such as prompt writers.  We think the same will happen with fundraising, with more mundane work being taken away and fundraisers having more time to concentrate on more interesting and valuable work. 

For example, our new bid writer can already create as many grant applications as anyone wants 24/7, and we'll continue to develop it and the GPT software will also continue to improve.  However, we don't think it could ever have the important depth of personal experience a good bid writer has.  It's mainly aimed at non-professionals, although we think that it may well be used by fundraisers to very quickly create drafts for them to then improve on.  But we think there’s far more potential.  One area we're exploring is fundraisers who work with multiple charities/charity teams, having them use the bid writer to obtain the information required and then create an initial draft for the fundraiser.  It doesn't replace the fundraiser either talking to the charity/team to help them refine their thinking, or in the crafting a compelling bid, but it has the potential to lose a lot of the more mundane aspects of their role. 

Inevitably, it will be essential to support our people and provide them with skills development, but we see AI as enabling them to achieve even more of the fantastic work they do, not a way to cut jobs to deliver cost savings.  

The one area where we do see potential job losses, is charities whose work or business model is significantly impacted by AI, but who fail to respond.  Not doing so may result in loss of income, or beneficiaries simply going elsewhere, because they can get what they need more easily, and/or at no cost and/or delivered better.   See our questions to ask yourself at the bottom.  

To access our AI services, click the tech bunny icon in the bottom right of any screen and ask it short questions, including key words.  Our services work for any of the 0.5 million UK non profits and the estimated 14 million charity people.  These are free, available 24/7 and can handle an almost unlimited number of tasks.  Currently we have 4 services that:

We launched our first AI service at the beginning of 2023, 2 upgrades since and launched our first GPT (the ChatGPT software) service at the end of March 2023. 

It's a tool that enables anyone to create funding bids very quickly and simply.  You don't need to know anything about ChatGPT, or how to structure or what to include in a funding bid and you don't need to be good at writing either.  We've also designed out well known GPT quirks, such as hallucinating.  But, you get back what you put in to it, so it's a tool anyone can use, but it's not a magic button to press.

The tech giants are pouring £s billions into AI development and embedding it into existing systems, including browsers and even Excel.  By this time next year, quite a few of the tech issues will have been dealt with and pretty much all of us will be using it in one way or another.  However, it will take years for regulatory systems to catch up with the legal and ethical issues presented by generative AI.    

It will massively change our world, but whether it's an opportunity or a threat depends largely on what you and your charity do about it. 

I think those at most risk will be charities that choose to cling to services and/or business models that people won't need any longer, or they can access for free or more effectively or more easily using the new AI systems.  The risk may be greater for those that are larger and national, because they are less likley to have very niche or place based activities that are too difficult or small to make it worth building AI for.  

These will create huge change and very soon but, as with personal computers and Windows, nobody yet knows what those changes will be. Here are my suggested risk indicators you may wish to think about: