How Auditors are Helped and Hindered Advancing AI in the Workplace

Since OpenAI’s ChatGPT burst onto the scene in late 2022 and picked up unprecedented user growth in early 2023, speculation and hypothesizing about the future of finance and accounting with highly advanced large language models has been just as fast moving. Outwardly, accounting firms large and small have embraced the advance of AI and machine learning and spoken positively about the effect it will have on both their own and their client’s workspaces. 

‍

Internally however, there was understandable concern about data privacy and confidentiality. Just as staff were beginning to draft deliverables and write macros with ChatGPT, there were inevitable restrictions or outright bans placed on the use of OpenAI’s breakthrough chatbot. While PwC did not ban staff from accessing the platform, they did reiterate it was not to be used for client work. KPMG, Marcum and others meanwhile launched their own in-house LLM. Beyond accounting firms, up to 75% of companies surveyed have or are considering bans on applications like ChatGPT, with major tech leaders like Apple and Amazon among them.

‍

A year on from the hype surrounding the launch and growth of ChatGPT, we at Tellen wanted to speak to auditors and hear firsthand about how AI has changed their workspaces. With as much as a full audit year under their belt with ChatGPT (or as we found, often a proprietary chatbot contained within the firm), we sat down with auditors and got their thoughts on the impacts these technologies have had on their audits. We wanted to find out where they see AI being used currently, where it could be in the future and how attitudes toward the use of AI have evolved.

‍

When ChatGPT first gained substantial momentum there was curiosity and excitement within the rank and file of accounting firms about the potential to automate some of the more dreary elements of the audit cycle. While audit practices have never exactly been known to be on the cutting edge of new tech, this new development seemed to have a different flavor. New ideas were going to be experimented with and implemented from the bottom up, with the staff to middle management level potentially leading the charge equipped with democratized AI.

‍

One manager at a Big Four firm told us: “So much of managing the audit can be boiled down to administrative burden; following up with a client for late deliverables, coordinating with overseas teams and just making sure our audit is still moving”. She went on to say “Everyone offers the advice as a new manager to ‘set aside 30 minutes for emails once or twice a day’ but that’s not nearly enough. Having ChatGPT write my emails for me, especially when I was too tired during busy season, was a real time saver”. It seemed at least initially that AI might be the tool that gave auditors the extra time or energy they needed in their day, with potential to expand into an even more integrated and efficient tool. 

‍

Auditors told us, however, the lack of ability to integrate chatbots into their existing systems was one of the main reasons they couldn’t go beyond the surface level in their use of the tool. “It would be great if I could ask the chatbot to draft up an outstanding items list and sent it to the client but unfortunately even if it could access and read our portal, the firm would never allow it” the same audit manager told us, which also speaks to the question of not only practically how but if such tools will be allowed to access client data. The interface problem isn’t necessarily solved by the introduction of internal chatbots launched by some of the large firms, either. One senior associate at another Big Four, which has launched its own LLM, tells us the tool can pull on a bit more firm specific knowledge but it can’t integrate seamlessly with their audit tools, meaning it is still essentially a chatbot where you can input and extract natural language but not much more. “I know people have used it to write proposals and reports but they haven't been able to automate large chunks of work like we were thinking at first”. 

‍

Outside of the largest accounting firms, industry is looking to capitalize on the use of AI, albeit usually without proprietary technology. “We just had our head of internal audit talk about using Gemini with a corporate license." a senior internal auditor at a mid-cap public company shared with us. “While we're not barred from using Gemini or ChatGPT, we're also in the process of developing an AI use policy, but the attitude toward using, while somewhat hesitant, has been positive overall.” 

‍

One common thread we noticed throughout all our interactions was that the initial excitement of using AI tools seems to have dissipated as firms get their arms around the technology, place restrictions on its use, and form specialized AI consultancy divisions. “It does feel at times like the audit practice is getting left behind a bit” the same audit manager told us. “It seems like we're focused on selling AI to our clients and not as much on using it internally to make our own work more efficient”. Beyond the hard and fast rules about whether and when AI can be used, we noticed a disconnect between some of the outward facing discussion of AI by firms, seeking them to position themselves as experts, and how new technologies are embraced in the audit practice. “I get the sense that they know this is going to be big but they’re still figuring it out just like we are '' said the same senior internal auditor. Whether due to lack of practical applicability or concerns about data security, accounting firms are navigating the waters of AI seemingly more externally than within their walls, at least initially. 

‍

While limitations of free chatbots like ChatGPT are an obstacle to further use of AI, it will take the full support of firms and initiatives focused on experimenting with using AI in real world and environments and to perform audit procedures. At Tellen, we’re striving to develop the AI tools that will drive efficiencies in all phases of the audit, as well as enhance quality by drawing on large data sets of quality findings. We believe that AI can not only absorb tasks that are ripe for automation but also enhance the insights and ingenuity of auditors. Our aim is to allow auditors to build these automations in natural language within the infrastructure we have created, with data security at the forefront. To learn more about Tellen, contact us via our about page or join our waitlist via the links above.

‍