INTHEBLACK September/October 2025 - Magazine - Page 43
“Rather than just having AI as a separate
tool, it needs to be permeating our work
practices, and we need to become familiar
with how it changes day-to-day work.”
MICHAEL DAVERN FCPA,
THE UNIVERSITY OF MELBOURNE
emails, but which ultimately is designed to
adopt the traits of a seasoned Goldman Sachs
employee. The significance of this development,
according to Salampasis, is that it combines AI
and behavioural science, rather than just focusing
on productivity.
HARNESSING VALUE
A Boston Consulting Group (BCG) survey of
C-suite executives shows that business leaders
really want to generate tangible results from AI,
with the focus being on reshaping key functions.
The report also indicates that many companies
have been slow to adopt AI training, while they
often dilute their AI efforts by pursuing too many
use cases.
BCG advocates three key ways to maximise
AI potential: deploy AI in everyday tasks to realise
10–20 per cent productivity potential; reshape
critical functions for 30–50 per cent enhancement
in efficiency and effectiveness; and invent new
products and services to build a long-term
competitive advantage.
The survey reveals two-thirds of companies
are exploring the use of AI agents — that is,
advanced systems as part of agentic AI that
can act without the input of humans.
Julian King, partner and director at BCG X,
a tech build and design unit of BCG, says
agentic AI is proving invaluable in the discovery
process as government departments and banks
seek to modernise legacy computer systems.
“Typically, [this process is] very laborious and
takes an army of people,” says King, who adds
that the new AI-inspired approach can deliver up
to 300 times the productivity gains on some tasks.
However, the BCG research reveals that 60 per cent
of companies surveyed are failing to define and
monitor any financial KPIs related to AI value creation.
BCG X managing director and partner Kevin Lucas
says a failure to set robust AI policies or KPIs can
result in companies “letting a thousand flowers
bloom” and then hoping that some survive and thrive.
“What you see, therefore, are lots of teams building
really small AI use cases in an uncoordinated way.
So, the returns tend not to be significant.”
Best practice is for business units to set
budgets with specific financial targets to reduce
operating costs.
Professor Michael Davern FCPA, chair of
accounting and business information systems
at The University of Melbourne and a member
of CPA Australia’s Digital Transformation Centre of
Excellence, believes AI’s capacity to reshape work
is of greatest value — in the short term at least.
“Rather than just having AI as a separate tool,
it needs to be permeating our work practices,
and we need to become familiar with how it
changes day-to-day work,” he says.
Davern believes more training on prompt
engineering — the art of instructing Gen AI tools
to generate the desired output or information — is
desperately needed, along with the skills to analyse
and unpack that information. “Sometimes it’s just
not fit for purpose, and there are hallucinations,”
he says, a term that refers to the incorrect results
that AI models may generate.
WHAT ABOUT ACCOUNTING FIRMS?
AI can clearly help accounting firms automate routine
tasks and assist in areas such as cash-flow forecasting,
compliance checks and fraud detection.
GLOBAL GAME CHANGERS intheblack.cpaaustralia.com.au 43
