How to use generative AI to build a business case for generative AI

In August this year, PricewaterhouseCoopers announced that it would invest another £100 million in cutting edge technologies including artificial intelligence (AI).

As well as integrating new software into the business, the investment will include training all PwC staff in the United Kingdom in generative AI skills.

The Big Four firm will use generative AI to create, enhance, summarise and analyse unstructured data.

While machine intelligence is transforming professional services across the globe, those not investing in generative AI tools and training risk being left behind.

Not only is the development of staff at risk, but employees may decide to seek work at another firm that is committed to staying up to date with technology.

Reasons to invest in generative AI tools and training

AI can not only help accountants be more effective business partners to their clients, but help them keep SMEs afloat, suggests 2023 research by Intuit QuickBooks.

Forty-three per cent of accountants agreed that not adopting AI in the next five years will negatively affect business growth and others worried they would lose clients.

Conversely, accountants see the benefits of AI including automation of routine and repetitive tasks such as data entry and reconciling transactions, gaining time to focus on business advice, greater ability to spot data and future trends, and more accurate cash flow forecasting.

London accounting firm Tax Agility gives the example of a small restaurant thinking of expanding. Using AI-powered financial forecasting to analyse sales data, customer demographics and seasonal business, an accountant could offer advice on the optimal timing and location to open a new eatery.

Other benefits of generative AI include reducing human error, enhancing workforce skills, elevating staff training, tailored job descriptions and improved security through identification of data anomalies.

AI products such as massive data analytics are also useful when accountants are looking for exact solutions, at scale, in real time, IPA Group Director of Education Philomena Leung explained recently.

Using AI to grow: Tools and plugins to know now

AI-powered data/plug-ins are already being used to automate data entry and reconcile processes.

• Expense management: tools such as Expensify and Receipt Bank can automatically extract data from receipts and generate reports.
• Analyse complex financial data and identify potential deductions: TaxJar and Avalara
• Visualise and interpret financial information: Tableau or Power BI
• Identify fraud or irregularities: MindBridge AI Auditor.
• Produce detailed reports and financial analysis: GPT4

GPT4, OpenAI’s latest release of ChatGPT, accepts inputs of both texts and images. A range of plugins mean that spreadsheets can be uploaded and analysed, and GPT4 can identify insights in data, as well as share those data insights in graphs and copy.

While it’s not a replacement for human logic – it strings words together without actually understanding the ideas it expresses – it has been shown to perform at a human level on some professional and academic benchmarks (like passing the bar).

The risks of generative AI

Generative AI models aren’t perfect. The risks generally fall into two buckets – the risk of exposing data by sharing it with an AI tool, and the risks associated with using inaccurate information provided by the AI tool.

In the first bucket, the risks of inadvertently exposing data by sharing it with a public tool are best mitigated by a very simple rule: if you would not like to see something published publicly, don’t share it with an open AI tool such as ChatGPT.

For internal and paid AI tools:

• Create detailed guidelines about what data team members can upload.
• Set sensible access restrictions so that only the people who need access to data and the tools have access.
• Ensure team members all use two-factor authentication to protect their logins.
• Only use client data in ways that clients have approved.

In the second bucket we have all the risks associated with AI lying. Large language models such as GPT4 hallucinate – or make up facts, quotes, data and stats – because they don’t understand the meaning of the words they produce. They are stringing together words without comprehending the ideas those words express.

The risks to users are not small, for example citing regulations incorrectly and mistakenly advising a client to claim a benefit.

Checking the insights produced by AI tools can be time consuming, but is essential. Fact checking should be detailed – one strength of a tool like GPT4 is its ability to mimic language, so a paragraph quoting a regulatory guide, for example may sound spot on, but be entirely fictitious.

How to use generative AI to create a business case for generative AI

If you are keen to use AI tools more, but need to make a case for the business to invest in training and subscriptions, try using simple and accessible AI tools to streamline the creation of a business case.

• Create copy: Ask ChatGPT to write a business case for integrating AI tools into an accounting practice in [x location], with [x] team members, customers in [x and x] industries, and annual revenue of [£x]. In the chat thread, keep asking ChatGPT to refine the business case until you have a solid draft.
• Create a presentation deck: Open SlidesGo, Gamma or Canva, and generate a presentation deck. You can choose a design and add your topic, and these tools will create a basic structure, and even add words.
• Update the presentation deck to suit your company’s brand, and add your ChatGPT-generated copy. As you go through, you’ll edit the copy and check the facts. You can go back to ChatGPT for any missing information.

It won’t be perfect at this point, but it will give you an enormous headstart.