Risks loom even as artificial intelligence holds promise of productivity for ASX companies

ASX companies are joining the AI race at full throttle, but caution is required because the technology has flaws and there are significant risks about misinformation being conveyed.

Artificial intelligence is the new frontier and must be assessed on its shortcomings as much as its benefits.

Artificial intelligence holds out the promise of transformation and productivity gains across the ASX, but coupled with that are real risks.

It’s a topic that is top of mind across a range of sectors, particularly as the technology evolves and generative AI – which can create content such as text, images and videos – gains traction.

Caution is required, though, given that the technology has flaws and there is a real risk of misinformation being conveyed.

US Treasury secretary Janet Yellen put it bluntly last month when she said there were “tremendous opportunities and significant risks” associated with financial institutions using AI. Because of that, the issue has gravitated towards the top of the US Treasury’s and Financial ­Stability Oversight Council’s agendas.

Treasury Secretary Janet Yellen has acknowledged the opportunities posed by AI but warned of the risks. Picture: Getty Images

Even the McKinsey Global Institute – which estimates that across the global banking sector generative AI could add between $US200bn ($296bn) and $US340bn in value annually, via increased productivity – is warning of risks. Among them is the generation of false or illogical information, intellectual property infringement, limited transparency in how systems operate, issues of bias and fairness, and ­security concerns.

To put it to the test, this columnist asked ChatGPT to identify Australia’s largest ever takeover transaction. Its answer was that one of the nation’s biggest takeovers was SABMiller’s $12.3bn acquisition of Foster’s in 2011. While that is technically correct, it is not the question I asked and that deal has been well and truly surpassed as Australia’s largest acquisition.

When I asked again whether that was really the biggest takeover, the response changed and the AI apologised, citing BHP’s tilt for Rio Tinto – once worth $140bn – as the answer, despite the deal being withdrawn.

In fact, the largest completed acquisition of an Australian target was the purchase of Afterpay by US company Block, valued at $39bn on announcement in 2021. That is based on London Stock Exchange data.

That is just one example, but it illustrates the caution that’s required by ASX-listed companies as they assess the merits of drawing on AI, even for relatively mundane tasks.

There will be winners and losers in the AI race.

Broking house Morgan Stanley recently identified 10 Australian listed companies for which they deemed AI adoption was progressing enough to be “relevant to the investment case”.

The list includes some understandable stocks and a few unexpected names. The 10 were: WiseTech, REA Group, Macquarie Group, Coles, Woolworths, National Australia Bank, Life 360, Sonic Healthcare, Transurban and Origin Energy.

The research noted the current technology shift was progressing “faster than any before it”.

“Companies that best adopt this new technology to scale their businesses while maintaining robust barriers to entry accrue most value,” Morgan Stanley said. “As AI becomes more core to the investment thesis on any stock, it would likely lead to more significant re-ratings and de-ratings for beneficiaries and disrupted companies, respectively.”

For pathology group Sonic Healthcare, the broker thinks AI adoption could drive “material upside” to its earnings expectations from factors including cost efficiencies, if the company can streamline pathologist workflow.

Another potential positive is increased revenue from licensing out products from its joint venture with Harrison.ai, which is aimed at speeding up the detection and diagnosis of cancer.

For Coles and Woolworths, Morgan Stanley said they were major adopters of AI to drive growth and cost efficiencies, but the broker posits that benefits will be partially “competed away”.

In that sector, AI algorithms can identify changes in consumer buying patterns and suggest prices for products that maximise sales, which would of course benefit the grocery giants.

In banking, the AI proposition is more nuanced partly because of the nature of the industry and the regulation imposed on it. Morgan Stanley reckons National Australia Bank may use AI to narrow the return-on-equity gap between it and Commonwealth Bank.

“While CBA has maintained its digital leadership and looks to be ahead of the other banks on AI, we believe NAB now has strong technology foundations in place, is well positioned to leverage advances in AI, and has more than 20 live GenAI use cases,” the broker said.

In a recent LinkedIn post, NAB’s operating chief, Les Matheson, said the bank was building a secure platform that would deliver the “infrastructure and guardrails” to test and scale generative AI.

NAB is assessing it in areas including bankers being able to search for information across applications, personalising customer marketing messages and helping the bank’s lawyers with trust deed reviews.

NAB has partnered with Accenture on its projects and the latter’s global CEO, Julie Sweet, told her investors NAB’s generative AI platform included the creation of 200 generative AI use cases in backlog, with 20 being tested across the bank.

However, CBA isn’t resting on its laurels. Customer engagement and digital executive general manager Meg Bonighton told this columnist there were thousands of potential AI use cases across the bank, but it wasn’t forging into AI for the sake of it.

For its banking app, CBA has drawn on AI to make the homepage more unique to the customer, and algorithms are being used in the background to determine navigation tiles, which can change based on how someone is using the app. AI is also drawn on in compiling content and prompts to send to customers, which has led to an increase in app engagement.

“The other space in the app that it (AI) is used is in the money management tools,” Bonighton said.

CBA’s interim results noted that “responsible scaling of AI” resulted in 50-plus generative AI use cases spanning the simplifying of processes and providing support to frontline staff.

AI is certainly an interesting area, but one that needs the involvement of risk management executives too.

As investors closely assess how the AI race will play out, they need to ask tough questions about which suppliers ASX-listed companies are drawing on for AI and what operational risks firms may be exposed to. The risks around the quality of information and data, and the potential for AI manipulation, are just a few.

Originally published as Risks loom even as artificial intelligence holds promise of productivity for ASX companies