When I tell people I’m an auditor, most people make a joke about me using a green pen (apparently auditors in banks used green pens so everyone knew an auditor had ticked/marked a document).
Fast forward 20+ years and the world of the auditor is now completely different. Gone is the day of the old part-time auditor (mostly a tax professional who dabbled in audit). They have been replaced by young, intelligent, career-oriented people, with a unique skill set – capable of dealing with big data, IT savvy, with strong interpersonal skills and the ability to challenge and be skeptical. These auditors need to combine technical issues with judgment and skepticism, and are able to balance attention to detail whilst considering the big picture of the financial report as a whole (sometimes referred to as Materiality).
Could your audit be performed by a robot by 2025?
Fast forward even more and there’s talk that auditors could – in effect – be replaced by robots. The World Economic Forum undertook a “Technological Tipping Points” survey in 2015 when 800 executives were asked when they thought that “30 percent of corporate audits would be performed by AI (artificial intelligence).” Seventy-five percent of the respondents thought that this particular tipping point would be reached by 2025. Somewhat biased, I beg to differ.
Software can perform a lot of the mundane tasks of an audit, such as checking debits and credits, authorisations and manual work previously handed to the poor newbie auditor (I recall my first few years as an auditor and shiver!). But how far will software take us though? AI is able to assist us, but it’s driven by code written by programmers. An audit is based on a skeptical mindset, meaning that we challenge what is put to us.
The limitations of AI software
We need to consider the strength and reliability of the audit evidence (verbal, written, external/internal etc), and its relevance to the issue. Only a human can think through how certain issues, disclosures and financial statements have been created. Are relevant accounting policies appropriate to the entity, what isn’t included in the financials, do warranties and provisions appear adequate, could there be fraud, are the assumptions reasonable, do the numbers comply with the relevant accounting standard? And so on.
I therefore wonder how far software will venture into these subjective, judgment-based areas.
Sure, software can grab data feeds and check that income, franking credits etc match those in the bank feed…but can it determine if a transaction is at arms-length, whether related parties exist, or whether the valuation of goodwill is appropriate? How will software tell us what isn’t there?
Qualities that the future auditor will need
The Chartered Accountants Australia and New Zealand recently released a short video on the future auditor, whose qualities include being a watchdog, professionally skeptical, independent, ethical, a problem solver, motivator, cross cultural, linguist, strategic thinker, innovator, mediator, story teller and juggler! This just shows us how the world of the auditor has changed.
I don’t see AI as the death of the auditor. (I’ve never seen a computer juggle…….yet!) But I do see it reducing the number of auditors, and pushing many outside their comfort zone. Those who thrive will be the ones who actively embrace the potential of AI, rather than shying away from the uncertainties of this new technology.
2 Responses to “Can artificial intelligence replace an auditor?”
I am working on one of Big 4 where I am in charge of IT Risk Assurance. Now I am investigating “Future Audit” including “AI”. I think that it is more important for today’s audit to assure the information generating and processing.If we will use Robots or the more intelligent computer,we have to investigate its processing accuracy and completeness much more than now. I think CPA must play the more important role and closely work with IT specialist or IT Auditor.
Very good article. Regarding the “subjective” areas of audits, AI with extremely large amount of past and comparable data would be able to do much better analysis on these areas compared with the educated guesses of our auditing friends. Subjective parts of the audit would become more objective as it was meant to be in the first place.