On September 27, 2023, Professor Michael Davern FCPA, Professor of Accounting and Business Information Systems, Department of Accounting, University of Melbourne, presented the 84th CPA Australia–University of Melbourne Annual Research Lecture.
This is the oldest continuous lecture series in the University’s calendar and the world’s longest-running annual lecture in the accounting discipline. Departing from the COVID-impacted formats of the previous three years, the lecture was delivered live as well as a webinar.
The formalities commenced with the chair for the evening, Professor Vic Naiker, Deputy Head of the Department of Accounting, drawing attention to the history of the lecture before introducing the speaker and outlining his distinguished scholarly background.
The theme of the lecture was the question of whether developments in artificial intelligence (AI), analytics and automation had created an existential crisis for accounting and accountants. In answering this question with a resounding “No ”, Professor Davern ranged over several themes and findings, including:
- The nature of AI, how it is being improved and the ensuing problems and challenges it creates.
- The features of professional judgment, how it is created and its challenges and limitations.
- The nature and challenges of augmented intelligence.
- Research findings into AI-type applications.
Improvements to AI take the form of additional processing power, greater data availability, faster activity in relation to decision making and increased automation of data collection and decision making. However, AI’s efficacy remains dependent on the quality and veracity of its input data. The problem of garbage-in causing garbage-out has not been obviated by AI. Accounting expertise has to be focused inter alia on designing business processes to ensure data veracity and then to interpreting any patterns in the data identified by AI.
While a necessary condition of AI applications is the existence of relevant data “histories” that facilitate the identification of patterns and relationships, there will always be situations lacking such histories. In these circumstances, AI is irrelevant and accounting advice on innovation and adaption will have to be based on human judgment, not machine learning.
Overall, AI has not removed the need for accountants to exercise professional judgment, it has simply altered the contexts in which judgment was applied. As to the characteristics of judgment itself, three different perspectives—ethical, cognitive and business—were summarised. For the former, the following definition of the IESBA was cited with approval:
Professional judgment involves the application of relevant training, professional knowledge, skill and experience commensurate with the facts and circumstances, taking into account the nature and scope of the particular professional activities, and the interests and relationships involved.
The cognitive perspective drew on psychological insights into human behaviour including the concepts of bounded rationality, reliance on heuristics and the existence and development of biases. A crucial “human ” feature was the creation of narratives and stories in relation to data. The business perspective was intensely practical; judgment was seen as having the potential to create value for business and society.
The second half of the lecture was devoted to five case studies of applications of AI. The first, a field experiment at Boston Consulting Group, was on the extent to which AI enhanced human performance. The remaining investigations had all involved the speaker, either as a PhD supervisor or researcher, with the first two focussed on augmented intelligence: the combination of AI and professional judgment. These studies involved, respectively, AI decisions in a marginal credit-risk situation, and demand management at two hotel chains. The next two studies concerned data usage: firstly, at board level, then as an initiative-enabler. The final study was into how best to redesign automation processes to leverage human capital. Among the key findings were:
- There is a “frontier” to AI’s capabilities. Beyond this frontier, AI is less likely to produce correct solutions than decisions based on professional judgment.
- Ideally, machine reasoning should be transparent and aligned with users’ incentives.
- AI decisions are enhanced when users have “interactivity” – the ability to explore adjacent decisions.
- The success of data-driven organisational initiatives is influenced by analytics education, and management and community engagement.
In his concluding comments, Professor Davern looked to the future of both AI and professional judgement. AI will become increasingly invisible through being embedded in applications that will change, but not replace, human work. Challenges will remain in relation to data quality, potentially compromised ethics and resource consumption by AI applications.
In relation to professional judgment at the personal level, accountants must recapture their role as trusted advisors while becoming better data “storytellers” and enhancing their expertise in the controls and risks of business-process design.
From a broader external perspective, audit and assurance will become increasingly important, particularly in assessing controls over process- and data-integrity, while standard-setters and regulators will be challenged by the need to allow greater use of AI and professional judgement in enhancing accountability.
Following the presentation, Professor Naiker was the conduit for a lively Q&A session, largely focused on skill development in the accounting curriculum.
In closing proceedings, Mr Ted Turner, President of the Victorian Division of CPA Australia, emphasised the rich tradition of the lecture as well as complimenting the speaker on the presentation.