AI is changing the work. Accountants still carry the accountability.
As AI continues to transform the accounting landscape, the crucial issue of accountability remains. While AI excels in handling routine tasks, the responsibility for accuracy and compliance still lies with human professionals.
Part of our AI in accounting coverage. See the full AI for accountants guide →
A question raised in an online discussion in the Australian accounting community last week drew dozens of replies. The member who asked said clients and people at networking events had raised it with them. Some had said they would not let their children study accounting.
The replies split in familiar ways. Some practitioners pointed out that they had heard the same prediction about computers, cloud software and bank feeds. Others said AI is already taking over routine work and will take much more of it.
Across the discussion, one point came up again and again. Someone has to be accountable for the work.
What practitioners are seeing
Many replies agreed that processing, research and repetitive tasks are where AI fits best. Several said they would welcome AI taking on the processing and routine work.
Some shared what they had seen first-hand. One accountant described a client who had moved bank reconciliation and bill processing to an AI tool. The tool emails the accountant to confirm any issues it finds, and the accountant reviews its work. Another practitioner asked how a setup like that would handle GST on a receipt with mixed items.
Another practitioner said a client had handed over an AI-prepared wages reconciliation that applied backpay across different award levels. In that practitioner’s view it was a reasonable attempt, but it was not correct. Their point was that if the work still has to be checked, the saving is hard to find.
A BAS agent said a software provider had told them its product would be fully AI-driven by early 2027. They had raised concerns with the provider about errors the program was creating or missing.
Where the accountability sits
The most consistent theme in the discussion was liability. One bookkeeper said that if a business was audited, blaming the AI would not stand up, because AI on its own has no accountability. Another practitioner said a software company will never be the one that takes the blame. One commenter argued that if software vendors say their AI will do the work, they should carry the liability and the professional indemnity insurance that goes with it.
Several practitioners connected this to how the profession is regulated. A longer reply argued that only a registered practitioner can take legal ownership of a tax return or advice. In their view, a tool that charged a fee for tax advice would run into the same rules. Another pointed out that software vendors and AI tools do not have to follow the same rules practitioners do.
The ATO came up as well. One practitioner said that a self-lodged return with inconsistent workpapers, and nobody who can explain why a decision was made, is unlikely to go down well.
Some saw new work emerging from this. One practitioner suggested a new category of work correcting AI mistakes. The BAS agent who had raised concerns with the software provider expected roles to shift towards compliance and fixing errors.
Read together, these replies treat AI as a change to who prepares the work. The responsibility for it stays with the person who signs it off.
What the work becomes
Several of the most detailed replies focused on what AI cannot easily do. One practitioner wrote that even a perfectly prepared return depends on whether the agent has the full picture. Clients often don’t know which details matter until someone asks. A short conversation can surface something that changes the advice completely.
The same practitioner described responding to an ATO query. AI can help organise the records and draft a reply. The agent still has to go back to the client, establish what actually happened and make sure the response reflects what the records support.
A shorter reply made the same point. The value lies in asking the right question and understanding the answer that comes back.
Others focused on what clients will keep wanting: relationships, trust and someone who understands how their business and family are connected. Some noted that fewer people are studying accounting, and that the shortage of bookkeepers is likely to grow as experienced practitioners retire.
On the original question about children studying accounting, the most practical advice in the discussion was to learn AI well. It also suggested spending just as much time learning how to talk with clients, spot what is missing and make sound decisions when the answer is not obvious.
What accountants can do now
The discussion also pointed to practical steps accountants can take now.
• Offer review as part of the service. Practitioners said AI will always need human review, and one was already checking an AI tool’s work for a client.
• Keep a clear record of how decisions were made. Note why each call was made, so the work can be explained if the ATO asks.
• Ask the questions that change the advice. Make time in client conversations to find out what the records don’t show.
• Build AI skills alongside client skills. Give both equal time, as the discussion suggested for anyone starting out in accounting.
• Use AI within privacy and legal limits. One practitioner suggested using AI within privacy and legal constraints. Another raised concerns about people pasting tax file numbers and financial statements into public AI tools.
Where the line sits
The discussion left the prediction open. Practitioners disagreed on how much of the work AI will take and how quickly. Where replies raised accountability, most pointed the same way. The person whose name is on the work has to be able to explain it, and that is where accountants bring something AI cannot.
Get the editable AI policy template Download the AI policy template
The Word document (.docx), ready to fill in. Your profile unlocks every template on The Firm.
Or build it with AI
Open Claude or ChatGPT with the prompt filled in. It writes the finished document for your firm straight away, then asks a few questions to tailor it. No client data needed.
Works without client data. Don't paste client names, tax file numbers or ledgers into an AI tool your firm hasn't approved for them. AI data safety playbook.