Skip to the content


Finding value in AI

Artificial intelligence

By Ian Lavis on behalf of Praxity Global Alliance

Artificial intelligence (AI) is supposed to be driving an audit revolution but the pace of change is slow. How can auditors overcome the challenges to achieve practical benefits?

Employing robots and smart technology to access vast amounts of client data faster and smarter than their human counterparts is an exciting if potentially hazardous proposition.

Intelligence-based technology has the power to streamline accounting processes, speed up mundane work and free-up employees to provide better client service.

In its simplest form, AI is typically used in audit to look at data sets and formulate rules and classifications. Humans then provide validation to understand why the AI tool would identify a certain transaction to be fraudulent or high risk.

Ignore it at your peril, warns Patrick Morrell, Chief Revenue Officer of Anduin, an AI start-up which has developed an intelligence-based billing platform. In an article published by Accounting Today earlier this year, he says failing to act on AI was once permissible but is now an “existential threat”.

But unlocking the value of AI is no easy task, partly because the use of technology to process vast amounts of data throws up difficult questions over ethics and confidentiality.


Barriers to adoption

Adopting AI poses significant challenges for accounting firms, especially in terms of data privacy legislation and client trust.

The majority of accounting professionals in independent accounting firms are still finding their way. Travis Webb, National Director of Assurance Services at BKD, one of the firms looking at how best to unlock the potential of AI and other intelligence-based technologies, says: “One of the biggest challenges we face is protecting confidentiality of client data. In the case of audit, even though we had access to ledgers, we did not have authorisation to share content.”

The US has particularly strict data privacy laws which impact on how data can be leveraged to train and then use AI. This could seriously hamper attempts to adopt AI technologies to improve accounting processes.  These tools require historical data to learn from – stacks of data that teach the computer what is and what isn’t an appropriate transaction.

Some firms have incorporated language in their contracts asking for permission to use general ledger data to build AI tools. However, while clients have been willing to allow a certain degree of content sharing, this effort has to date been on a small scale.

In jurisdictions where privacy legislation is more onerous, the benefits of AI will be harder to come by. Commenting on the likely way forward, Travis says: “I do not anticipate a quick and broad impact from AI in the US. Instead, it will be in directed areas where data is available.  With currently available technology, I do not reasonably see most firms building their own AI capability.”

There are also question marks over the use of AI when professional judgement is required. Bill Armstrong, Chief Innovation Officer at US accounting firm Moss Adams, explains: “One of the more challenging hurdles to AI solution adoption is the commingling of professional judgements with routine processing or data transformations. When you deliberately isolate those routine processes and tasks, you can successfully build robust AI automation around them. The ability to capture professional judgement in an AI solution is still out on the horizon somewhere.”

A report by the AICPA entitled The Data-Driven Audit: How Automation and AI are Changing the Audit and the Role of the Auditor cites trust as a potential barrier to AI adoption, especially when processing more complex data.

Trust issues arise when the data points become too complex for the algorithms and AI tools to readily document the cause-effect pattern, the AICPA report states. In this case, advanced AI techniques use a series of algorithms to recognise relationships between large volumes of data which may be difficult to understand or document, resulting in “lack of transparency or explainability of sophisticated AI tools”.

An ongoing challenge is the difficulty in achieving the right balance between technology and human resources. One of the early arguments against wide-scale AI adoption was the possibility of wide-scale redundancies and disruption to staffing, particularly at data-entry level.  

There now appears to be a growing acceptance that firms would be better placed to serve clients when technologies address the routine work and staff are employed at higher advisory levels. It could be argued the transition could actually create new job opportunities in the building of AI solutions.


Where is the value?

Despite the considerable challenges of adopting AI and related technologies, there are plenty of practical options on the table to improve businesses processes and client services.

To date, the focus has been mainly on the use of automation and analytics rather than full-blown intelligence-based technologies.

The AICPA report states: “For many auditors, using automation and analytics is a first step in their digital journey towards an AI-enabled audit. Much like the digital advancements that preceded it, AI will perform repetitive tasks, provide greater insights and improve efficiencies and quality, allowing auditors to better use their skills, knowledge and professional judgment.”

However, firms need to be realistic about taking this first step towards an AI-enabled audit, according to the Moss Adams innovation specialist. “I believe that there is a misconception that highly specialized AI solutions will drive overall cost efficiencies in the short term. Realistically, what AI offers is a solution to when work gets done. Building, customizing and refining AI solutions outside of normal business cycles to ensure effectiveness allows you to apply those models during crunch time with greater speed and accuracy,” Bill says.

He adds: “There is a new set of cost elements related to more highly skilled, cross trained professionals who can tune AI solutions to the specific tasks they are being deployed against. As we know from the market, highly talented public accounts are in high demand; the subset of those who understand machine learning is an even rarer find. In the short term this causes pressure on the ‘cost efficiencies’.”

Moss Adams, BKD and other independent accounting firms within Praxity Global Alliance are exploring how best to exploit AI technologies to achieve practical benefits.


New technologies being investigated include:

  • Machine learning contract analysis
  • Advanced risk and planning analytics
  • Robotic process automation
  • Optical character recognition technologies
  • Automated and intelligent data curation
  • Predictive analytics
  • Data visualization tools and techniques


Many firms are investing in training professionals so they can adapt to new ways of working. Bill explains: “At Moss Adams, we look at AI technology through the lens of our people. We focus on routine, high-volume processes that our professionals have to do, but have a lower value proposition to them and our clients. We then look to engage our professionals in skill development to meet the growing demand for this new brand of professional. The trivial tasks are removed from their job description and we add AI automation building and validation tasks.”

Among the technologies with the potential to add value relatively quickly is a powerful new software called MindBridge, which enables huge datasets to be processed quickly, and aids long and short-range planning and evaluation.

MindBridge software leverages AI technologies in mapping data and evaluating results. Travis explains: “By populating algorithms with expert information, the tool can identify transactions that are out of the ordinary. This could be of particular benefit to auditors but maybe also to internal accountants. We are actively engaged and working closely with MindBridge. We are collaborating on what would be useful or achievable.”

There is also significant interest in AI tools that simplify the process of dealing with complex regulatory guidance. International law firm Hogan Lovells, for example, has developed a European Banking Authority (EBA) outsourcing tool to help businesses tackle logistical challenges when dealing with EU regulator’s new guidelines.

This machine learning software has been trained to process outsourcing agreements and identify any terms that need to be updated to be compliant. The software is supplemented with human review quality control to ensure only relevant issues are sent to teams for varying and redrafting.

It is clear the practical benefits of AI and associated technologies are still very much in their infancy and, it has to be said, rather unexciting at the moment. Nonetheless, there is value to be had. “For those people that think AI will solve everything at once, there is a lot of lower hanging fruit to be tackled first to alleviate some relatively mundane tasks,” Travis says.


Getting started

For those firms just beginning their AI journey, knowing which path to take can be daunting, especially identifying machine-based ways to improve business processes that are both workable and affordable.

The AICPA recommends auditors follow a clear path to adopting AI technologies best suited to specific tasks. Here is a summary of the key steps:

  1. Get educated - learn about AI opportunities, how clients and other organizations are leveraging AI and how it can be applied in your firm
  2. Identify AI leaders within your organization – if there aren’t any, ask why not
  3. Identify opportunities for automation – focus on high-benefit, low-effort opportunities such as reviewing spreadsheets, filtering and sorting information, reviewing documents and data entry
  4. Identify opportunities for AI – focus on tasks that require you to look for patterns in data or opportunities for checking for patterns in high volumes of data that would be challenging or time consuming for a human to do;
  5. Get support from relevant accounting bodies.


Further reading:

The Data-Driven Audit: How Automation and AI are Changing the Audit and the Role of the Auditor