Safe and practical artificial intelligence tools for accountants

How are accountants using artificial intelligence tools?

A 2024 survey conducted by Karbon Magazine on a sample of accountancy firms saw 82% of respondents say they are intrigued or excited by AI, yet only 25% are investing in AI training. 

This stat tracks with our own experience on this topic. Talking to our accountancy partners, we discover a picture of an industry that's excited by the potential of this new technology, but unsure about its practical implementation.

If that sounds familiar, this blog post is for you!

We're Impelling, a Managed IT Services company who work with many clients in the professional services sector. The nature of our work means we have a responsibility to evaluate new technology, so we can provide solid guidance and recommendations to our customers, which is exactly what this blog post aims to do.

We'll take a look at how accountancy firms are using AI, what the security implications are, and discuss how you can help your staff and clients benefit from AI, safely and responsibly.

And no, this post is not written by AI, we're humans over here, so feel free to reach out if you'd like to chat!

What can accountants use AI for?

To answer this one, it's worth understanding what the different types of AI are, and how they fit into your working day.

At the time of writing, some of the big players in the industry are introducing AI tools into their accountancy packages. It's still early days, so most of these tools are focused around generating reports, using the AI to trawl through the data held in the platform and hopefully present it to you in a way that makes sense. These tools tend to be tightly integrated into the software, so using them doesn't feel much different than using the non-AI features of the software. It's just processing information behind the scenes in a different way.

The other side of the coin is chatbot style AI programs, like the very popular ChatGPT, or it's Google flavoured cousin Gemini. These act more like an assistant you can instruct to do tasks, requiring more input, and a bit of finesse to get results from.

The third - still quite experimental at the time of writing - side of this strange shaped coin is "computer use" AI. This is where you can instruct AI to take the driving seat, giving it control of your keyboard and mouse, and allowing it to carry out tasks on your screen. You can take a look at a demo here.

To answer the question then - what can accountants use AI for? AI tools integrated into software allow accountants to analyse data with a bit less effort. Maybe even generate some reports that give a unique perspective. Computer use AI has great potential for completing repetitive tasks like data entry. For everything else, there's the assistant chatbots, designed to be useful for a wide variety of tasks, from writing written reports to error checking spreadsheets.

Is it safe for accountants to use AI tools?

Most AI tools work by hoovering up as much data as they can in order to crunch it behind the scenes and spit out helpful information. This often includes the data you input, so needs approaching with care.

It's crucial to first be aware what the AI tool you're using has access to. When Google turned on the Gemini features in Google Workspace, some organisations got a shock when they found out their file sharing permissions weren't strict enough. Given the machine's ability to poke in to every nook and cranny of the shared drive, users found they were able to ask it things like "how much does everybody get paid here", and the robot would diligently deliver.

There's also been cases where people have stress tested tools like ChatGPT, and through a series of strange prompts, managed to get it to output genuine personal information, presumably previously hoovered up at some point.

The concern here is the sensitive financial information the people in the accounting industry work with on a daily basis. While you might be able to upload a payroll spreadsheet to ChatGPT and have it complete a job in the fraction of the time it would take to do it by hand, the question about the risk of doing so needs to be asked.

We advise checking the privacy policy, and data protection policies before using any AI tool, and making sure that you're not unwittingly disclosing the sensitive information, or trade secrets of your customers.

Understanding the limitations of AI

AI, and in particular, generative AI tools like ChatGPT do come with a set of limitations that are important to be aware of.

While the models powering the tools get better with each generation, accuracy of information is the main limiting factor when using them. Given a task, the AI is programmed to help as best as it possibly can, and in some cases this can lead to mistakes or even made up information.

For workers using AI tools, this means it's essential to double check the output, especially when you're working with anything as critical as financial data. For some tasks, that checking process can often take just as long as doing the task yourself, undermining the usefulness of the tools all together.

For less crucial tasks, there's often still a large amount of remedial work to do. Let's say you're generating a blog post for example. Writing these things takes time, so giving a chatbot a list of bullet points and asking it to expand upon them is a tempting proposition. Again though, the content it produces needs checking. When filling in blanks the tools have a tendency to make things up, and it's not uncommon for them to offer products or services that don't exist. In addition, there's a tendency to swing towards US spelling and grammar conventions, which again can be time consuming to check and edit.

Using the tools is very much a personal endeavour, and how useful they are is going to vary between people.

What's the best AI tool for accounting?

Our particular AI tool of choice for accountants is Microsoft 365 Copilot. Copilot offers a standard chatbot type interface for generalised tasks, but also integrates into office apps like Excel, Word, and PowerPoint.

These integrations allow you to do things like generate advanced formulae, save time making reports, or reword written content. What we particularly like about these tools is that they're not a flat out replacement for the traditional software interface, they're more of a compliment. We find this shallows the learning curve involved with something like Microsoft Excel because users have a built in guide or assistant they can lean on when they need to complete an unfamiliar task. It'll be up to that user if they want to continue generating formulae with AI in the future, but as an educator we think AI very valuable.

Microsoft 365 Copilot also has a decent privacy policy, with Microsoft at least making the promise that your organisational data will be contained within the parameters of your organisation. 

If you'd like more of our thoughts on Microsoft 365 copilot, check out our review/write up here, or if you'd like to schedule a demo to see what it can do first hand, get in touch to book one in with a member of our team.

Final thoughts

The potential with AI tools is clear to see, but if you've stuck around reading until now, you'll understand there's a lot to consider. This probably explains the stat from the start of this blog post. Option paralysis, security considerations, responsible use, all contribute to inertia when it comes to AI for accountants.

As technology partners to our customers, we see it as our responsibility to help you get the very best out of technology, and given AI tools appear to be the future, we're investing heavily in strategies to help support our customers who want to benefit from it.

If you'd like to know more about our mission, please get in touch with us for an informal discussion about your needs.