8 Ways AI is Changing the Face of Accounting and Audit

By Danielle Supkis Cheek, VP, Head of Analytics & AI, Caseware

Over the past few years, the accounting profession has seen the rise of AI as a powerful enabler. It’s already created significant efficiencies in accounting processes, and firms of all sizes are using AI in various forms as an efficiency booster.

However, while AI can dramatically improve productivity in accounting, it also requires careful management and thoughtful implementation to minimize any potential for business risk. The key lies in understanding how AI can transform accounting, and how the profession can balance the benefits of AI with the need for appropriate safeguards and oversight.

In this article, we’ll take a look at eight considerations accountants should bear in mind as they adopt AI and integrate it to accelerate daily operations.

1. Take advantage of the democratization of AI technology

There’s a common misconception that AI is only for large, global accounting firms. The reality is that cloud technologies and AI-as-a-service models are making advanced AI capabilities more accessible to firms of all sizes. This levels the playing field for many accountants. These technologies reduce the upfront investment required for advanced AI capabilities, allowing smaller firms to access and implement AI solutions that were previously only available to those with substantial IT budgets. This democratization of AI is enabling a wider range of accounting professionals to leverage advanced technologies in their practice.

2. Understand that AI requires a well-thought-out approach to change

While AI may now be more widely available, it’s still a technology that needs a well-thought-out strategy surrounding both its implementation and ongoing use.

AI will require accounting professionals to think differently about their processes and be more iterative in their approach to change. As a whole, the profession will need to become more comfortable with iterative processes, similar to those used in software development. This involves testing, refining, and continuously improving AI implementations.

3. Develop methods for review and reliance on AI-generated outputs

Firms will want to consider how they supervise and review human work and how these approaches can be adapted for AI. In some contexts, it may be helpful to think of AI at the level of a third-year staff in terms of its understanding of complex issues and writing capabilities. This will dictate the appropriate levels of review.

It’s also important to understand the limitations and potential biases of AI systems to ensure their outputs can be relied upon appropriately. Even ‘black box’ AI systems can be tested using methods like flip tests, re-performance tests, and third-party assurance services. The latter are also becoming more widespread for AI systems, with specialized providers offering more advanced testing and validation of models, as well as more transparency across AI.

Read More : GlobalFintechSeries Interview with John Sun, CEO at Spring Labs

4. Understand the benefits of an incremental approach to AI adoption

Starting with small, manageable changes that can create immediate efficiencies is likely to be more effective than attempting large-scale transformations in AI. The incremental approach allows firms to build capacity over time, freeing up resources to invest in more substantial AI implementations when the time is right. It’s important to identify areas where technology can alleviate routine tasks, creating space for more strategic work.

5. Research how small language models may provide value

Small language models are an exciting development in AI technology that may offer the potential for more cost-effective and efficient deployment of AI in specific use cases. These models, derived from larger language models, may offer more efficient and cost-effective deployment of AI capabilities in certain scenarios, with reduced infrastructure costs and improved speed. Small language models could make AI more accessible and practical for a wider range of accounting applications, especially in situations where the full capabilities of large language models are not required.

6. Consider AI and accounting standards

Most accounting standards are technology-neutral, and the perception of them being barriers to AI adoption often comes from traditional interpretations rather than the standards themselves. When considering how AI will be implemented across a practice, accountants should revisit standards with a fresh perspective to ensure the approaches and processes surrounding the technology adhere to the principles laid out in the standards.

7. Be aware of emerging regulations

As AI becomes more prevalent, organizations will need to maintain an ‘AI inventory’ and be aware of emerging regulations like the EU AI Act. As AI becomes more prevalent in accounting firms, there will be a growing need for robust risk management strategies. There’s no time like the present for firms to create an AI inventory, to keep track of where and how they’re using AI systems, and where governance lies. This is the type of proactive approach to AI risk management that will be crucial for accounting firms to ensure compliance and maintain trust in their AI-enhanced services.

8. Aim high

Accounting firms have typically been heavily reliant on human review in their everyday work, but this actually makes them well-suited to leveraging GenAI.

What is interesting in terms of the rise of GenAI in accounting is the fact that existing processes for the public accountant are already highly reliant on degrees of human review and this wholly complements the recommended approach in terms of reliance on a GenAI output. Public accountant’s processes lend themselves to leveraging GenAI which will, in turn, reduce barriers and change management expenditure around the deployment of such technologies.

We’re already beginning to see a greater acceptance and willingness to accept AI as a mainstream technology in accounting, alongside a ‘humans in the loop’ strategy. Maintaining human judgment while leveraging AI will be the key to striking a balance between innovation and the profession’s established practices, and critical to leveraging the potential power of this compelling technology.