How ready are you to sell your business?

There have been vast improvements in artificial intelligence (AI) capabilities and availability this year. In this blog I look at the opportunities and risks that AI can bring for charities.

What is AI?

AI is a very wide-ranging term, but at its core, it’s a branch of technology designed to complete tasks which would normally require human input, done by simulating human intelligence.

Computers, though incredibly powerful, often require very precise instructions and formats of data to carry out very specific tasks which they have been coded to do.

AI shatters this limitation in a few ways:

• Using machine learning to ‘train’ itself by analysing what a person would do in the circumstance
• Having the ability to read and understand data in different formats and from different sources to compile an answer
• Using language modelling to be able to understand instructions without the need for precise inputs, and to be able to express an answer in a variety of formats or ways
• Recognising context and follow-up requests and context

How might charities use AI?

The potential uses are already staggering and with the development race ongoing between the main AI developers, the possibilities really do appear to be limitless. Some of the ways in which charities are already leveraging AI are:

• Automating written materials: For instance, grant applications, marketing materials, training information, internal policies.
• Reading and summarising information: Saving staff time by using AI to summarise long communications into more digestible levels of detail.
• Maximising fundraising campaigns: Analysing donor data to identify patterns and preferences and personalising communications to those donors to maximise engagement.
• Enhancing marketing decisions: For example, let’s say you want to know if there are any links between the timing of your social media posts and the number of donations it generates. Without AI you could probably do this to an extent by using a tracking link from the social media post. But what about the people who see your tweet on their phone and then go to their computer, search for your website and make a donation? As AI is capable of pattern recognition, it might be able to tell you that your donations are greatest when your fundraising campaigns post before the evening news rather than after.
• Providing information to the public: AI “chatbots” provide a quick and easy way for members of the public to get answers to questions about your organisation or the causes you support, perhaps leading to more donations, or reducing pressure on helpline staff.
• Speeding up reading: Converting text into formats which allow our brains to process it more quickly (such as “Bionic Reading”).
• Research: Whilst it isn’t entirely reliable (see below), it can often do most of the legwork in tracking down information and means a person needs only review it to ensure its accuracy.
• Automating repetitive tasks: Whereas previously it was possible to write a script to do this, it relied on two things: uniform data, and a computer programmer to write the script for you. AI can handle both, freeing up staff to add value elsewhere.

Things to watch out for

Despite its name, AI isn’t inherently ‘intelligent’. It processes vast amounts of data to provide a response, but, ultimately, it doesn’t know anything itself. If the data it’s pulling from is incorrect then its answer is almost certainly going to be wrong too.

It has been known not only to misunderstand information, but at times make it up altogether.

AI responses can be quite generic at times, with its language being clearly written by a computer model rather than a human being.

Perhaps most importantly, given the vast amounts of information being poured into online AI models such as ChatGPT, they are of great interest to data hackers. There have already been high profile data breaches where significant amounts of information (including user credentials) have been stolen.

Are these reasons not to use AI?

I don’t think so. The potential benefits are simply too great to ignore. However, as with any innovation, it doesn’t come without risk and it’s therefore essential to know and understand how AI is used and how the downsides can be mitigated.

The risks of incorrect information, or badly worded outputs, can be solved by making sure responses are checked before being used. Whilst not eliminating human intervention entirely, this converts the ‘doing’ into ‘reviewing’ which can still save a lot of time.

As for the data risk, much of this can be eliminated by leaving any identifying information out of the data you input. Names of donors, for example, might be redacted. Company names might be left out, to be added manually afterwards.

Where sensitive data does need to be included, usual guidance applies;  understanding what information is stored, how and where it’s stored, what security protocols are in place, etc.

Looking forward

We’re only just scratching the surface of what can be achieved with AI. There are certain technologies which arrive and which leave behind those who refuse to embrace them. I believe AI is one of those technologies and I, for one, am excited to see the amazing things which will be borne out of it.

Tom Dunston