Practical tips for using AI effectively
The introduction of Artificial Intelligence (AI) has become a key feature of the accountancy and professional services world in the past few years.
In the past six months alone, Deloitte Australia has been fined for providing a client with a report produced by AI and containing multiple errors, both Deloitte and PwC have moved towards job titles that focus on skills rather than titles in order to better reflect the use and emergence of AI, Grant Thornton have advised early career candidates not to use AI in their job applications and McKinsey have included AI collaboration as part of their early careers recruitment process.
It’s clear that AI will play a key role in accountancy and professional services in the coming years, so it’s crucial to better understand it and work with it effectively. With this in mind, here are some Accountancy Hub top tips on how to use it.
Technical tips
Before using AI or working on a project where it might be used, it’s worth spending some time familiarising yourself with your firm’s IT policy and guidance on AI-use. In particular look out for whether there are any limitations of using AI and if there are any associated disclaimers that should be used. Whilst this might not be the most exciting place to start, these are crucial first steps.
The golden rule - Only use your firm’s AI tools!
This is the big one! Most firms will have developed their own AI tools and environments, which have been developed to create a safe and controllable environment for data and will also be compliant with data-related regulation and legislation such as GDPR. In order to stay ethical and compliant you must ONLY use the tools approved in your firm’s IT policy.
Check the engagement contract
If thinking of using AI for a piece of client work, it’s probably worth checking the engagement contract, or asking the client lead if there are any AI-related clauses included in it. These days it’s likely there will be reference to AI use in the contract but it’s not unusual for clients to specifically request that AI isn’t used in work undertaken for them.
Check your disclaimers
The policy on this will vary between firms but some firms may require AI use to be declared in any client reports that are produced. If this is the case, keep track of when you’re using it and where the outputs from it are in the final report.
Practical tips
There are many ways in which AI can be used in practical applications. It’s a powerful, quick way of getting to an answer but it can also come with some risks, such as hallucinations (where it will make up information) and out-of-date information.
Check everything!
This one is crucial! You should be able to check and verify the information and outputs AI provides. This is important as it will not only provide you with an audit trail to refer back to if you’re questioned about it, but will also allow you to ensure the AI tool is drawing answers from credible sources. The best way to do this is to ask AI to provide sources to the answers it’s provided.
Using AI to edit
AI can be a useful editing tool, especially if you are concerned about the tone of what you’re writing or, more practically, if you are short on space in something like a pitch document. If using it in this way, make sure you use prompts that either tell AI what you want to do (e.g. “make this more concise”) or provide information on the audience for the piece of work (e.g. “turn this into bullet points that are suitable for a team update/Partner meeting”). Ensure that anything returned by AI, and used in a final output, reflects the tone and language of the rest of the work.
Using AI to do research
AI is a powerful research tool; you can ask it anything and, on the surface, it will return an answer that appears useful and credible. However, it’s likely there will be errors of some kind in the information returned. In order to mitigate against errors, there are some simple steps that can be taken. The first of which is to make sure you ask for sources to be provided in any answers (e.g. “Give me some information on Tesco’s strategy for 2026, sourcing and citing everything”)
Secondly, point the AI tool in the right direction by giving it the right documents to look at in the first place (again, important to reiterate that you should only be using the AI tools approved by your firm!). Several accountancy firms have developed their own in-house tools which allow you to upload documents, such as annual accounts, in order to be able to search for the most relevant information. This removes some of the margins of error, and when combined with a command to cite information too, can make for a powerful combination.
Finding the right formula
It’s worth remembering that AI is a new tool for everyone and people in the early stages of their careers may have greater experience of, and exposure to, it as a technology compared to more senior colleagues.
Part of the challenge in professional services is that there is a rush to digitise and to use data more effectively. This means greater emphasis on digital tools and dashboards, which mean many people are learning new data-related skills on the job.
AI can be a really useful tool to enable this as it allows users to troubleshootwhen they encounter a challenge. This might include asking AI to suggest a formula in Excel or suggest the next steps in data structuring for a PowerBI model. However, as always, be mindful of the data you’re uploading and only use the AI tools your firm is providing!
