Beware pitfalls while seizing AI opportunities

Speakers at an industry panel in Sydney last month urged caution as companies are preparing to leverage the innovative possibilities unleashed by artificial intelligence.

A Trojan horse dilemma: is generative AI a risk or an opportunity? A Trojan horse made of electronic circuit boards and OpenAI banners outside the Tel Aviv campus. Photo: AFP

-

-

Australian businesses are increasingly leveraging AI to increase their bottom lines and drive efficiency for customers – but executives are being urged to keep the technology accessible and transparent for end users.

Speaking at an industry panel in Sydney that was jointly hosted by The Australian and Nuix, several executives said they were moving quickly to capitalise on the benefits of AI but were also mindful of potential pitfalls inherent in the nascent technology.

Jonathan Rubinsztein, chief executive at ASX-listed software company Nuix, emphasised the synergy between AI and human empowerment, saying his company’s latest Neo platform could ingest and read up to 1000 different file types, leveraging on AI to process what was typically messy, unstructured data.

Nuix has pitched its AI-based Neo platform as a “force for good” for its clients, which range from large corporations to government agencies, legal firms and financial institutions.

“What it really does is it empowers a human that’s looking at data to be much more effective,” Mr Rubinsztein told the panel (part 2 of the panel discussion below and part 1 here).

“For example, we’ve just released our new data privacy product. We’ve trained the AI to say ‘What is sensitive data? What is a credit card? What is a social security number? What is a passport?’.

“When you’ve had a big cyber breach, the issue is that you typically have search strings and you search for ‘credit card’ and ‘passport’, and it would find a whole lot of false positives because it might find an email about a credit card that went missing. Then a human has to go and review that document.”

Nuix’s technology had reduced the number of false positives by up to 95 per cent, Mr Rubinsztein said, in what he described as an “explosion of opportunity” for his company.

He said much of Nuix’s AI functionality had been bolstered by its 2021 acquisition of the Boston-based natural language processing outfit Topos Labs.

“We’ve been very cautious as to how we build it because some of our AI is used by lawyers in the legal industry,” he said.

“So, that explainability is super important, because the lawyers will ask ‘How did you come up with this decision’, and the reality is that we still have the human making the decision. We empower the user to define the probability and the rating, and that accessibility is important. You don’t need a data scientist for that.”

Mr Rubinsztein said AI would ultimately improve outcomes for many people if harnessed correctly.

“I’m very optimistic about AI,” he said. “What happens is that technology enables humans to do a lot more. I’m super excited about the opportunity.

“I think there will be disruption in many industries, and people will need to relearn and figure out how to both leverage and harness AI, but I think it will change the world for the good.”

John Henderson, general partner at venture capital firm AirTree, said that the societal benefits from AI would likely be immense, as would the commercial opportunities. He was particularly excited about the implications for the healthcare industry, which he said would help augment the work done by doctors and nurses.

“You’ll be able to input a patient’s symptoms, and the software will give you a range of potential disease conditions,” he said.

“This is a perfect tool for a doctor to make sure that they are up-to-date, and that their diagnosis considers all possibilities and is validated by the latest science. The doctor can then focus more on being empathetic and patient-minded,” he said.

“So, there will be a bunch of diagnostic assistance tools that will lead to better health outcomes.

“This is true in the developed world where you have access to a doctor, and the impact will be even more exciting in developing regions and countries where no doctor is available.

“I think there’s a lot of exciting stuff to come.”

Simon Burns, partner at law firm Gilbert + Tobin responsible for its tech and digital practices, said it was critical that the federal government wasn’t too rigid in its approach to creating frameworks governing AI. The government is currently mulling its approach to regulating ethical AI as part of a month-long process.

“I think we have to be really targeted in understanding what is it that we’re trying to address and how best to do that given our existing regulatory framework,” he said.

“What is different about AI that requires a new approach, if a new approach for the relevant regulation is required at all?” Mr Burns said.

“And if we go to the question of guardrails, what are we guarding against and how do we most efficiently do that in a world where every bit of tech is going to have AI embedded in it in some shape or form? I think that’s the real challenge that no one’s quite grappled with perfectly.”

Professor Johanna Weaver, director of the Tech Policy Design Centre at ANU, said AI was already here. However, its ultimate outcomes – positive or negative – were still up for grabs.

“We, as a society, have a choice: we can actively engage to shape AI for the good, or we can naively expect it to be good. This is a choice that will shape our future,” she said.

“Accountability is important, particularly where AI systems are making decisions. People impacted should be able to understand the basis on which decisions are made. But it is not just technical considerations, it’s also about process. For example, can you appeal that decision?”

Professor Weaver pointed to the high-profile “robodebt” case as an example of a failure in transparency, accountability and process.

“Governments or companies seeking to embrace the productivity benefits of AI, must do so cognisant of the risk — this might sounds daunting, but tools exist to help,” she said.

“There are standards with handy checklists that set out, ‘if you’re building technology in this space, then make sure you have these things built in’.

“It’s about making sure that the people designing this incredible tech build to those standards. And ensuring that the people commissioning the tech are equipped to ask the right questions of the technologists.

“It’s cultural as much as technical.”

-

This content was produced in association with Nuix. Read our policy on commercial content here.

-