Published on the 27/02/2023 | Written by Heather Wright

It’s time for AI to go from low impact to big bang…

Andrew McPherson is adamant that AI can deliver Kiwi businesses real productivity improvements, today – and that it’s time to look to bigger impact deployments.

“Artificial intelligence is now ready for delivering value to businesses in serious customer facing applications and everyone should really be looking at what solutions are out there for their industry that they can deploy quickly,” McPherson tells iStart.

“It’s no longer just an experimental technology – something to be trying or looking at in the back room – it’s something we should be adopting and deploying now within our businesses.”

Stuff’s chief technology officer, McPherson will be talking at next week’s Digital First 2023 about the latest developments in AI and how they could change business and society.

“Companies can pick up these AI models, specifically trained for your industry, and apply them fairly quickly.”

“AI is going to have an impact on a wide range of industries, but there are concerns about its impact and, as we see at the moment, sometimes AI is just completely wrong – it’s certainly not at the stage where it’s always giving right answers.”

Deeper concerns also exist around the potential for bias and risks in terms of job displacement and the need for people to retrain, and the potential to exacerbate existing societal inequalities.

“One of the challenges is, knowing this ahead of time, how can we be more proactive and make sure we are being more inclusive with the technology right from the start?”

McPherson says while AI is being adopted in point solutions for specific industries, it’s often buried within an existing application or in data analytics.

“In enterprise AI there are a lot of very small point solutions or use cases where AI can be implemented but those may not actually have very big business impact. Often they’re more around data processing or maybe getting some sort of insight for the business that still has to be acted on by a human,” he says.

“I don’t think many people are using it for customer interaction or something that is customer facing yet.”

While customer service bots might be seen as a minor exception to that, he says the first step starts with looking across the business to identify manually intensive and repetitive processes, and then looking within your industry to see if there is an industry solution already developed and trained to solve the problem.

“At the moment, there is this base AI capability that has been developed internationally by some of the big companies like Google, OpenAI and AWS, and then companies are taking those capabilities and training the AI to industry-specific solutions. Companies like us are able to pick up these AI models that have been specifically trained for our industry and apply them fairly quickly.

“That’s why it is important to be across the trends and AI solutions coming out and look within your industry to see what’s been developed. It’s the quickest way of finding something that can deliver value,” he says.

De-risking the process also requires extended trials.

“If you’re looking at AI that will have a major impact on customer experience, or is customer facing, it really needs to be run in parallel with existing processes for an extended time to ensure you’ve tried and tested all the scenarios and understand how it will behave. First to de-risk the process, but also so you really understand what the AI is capable of and how it delivers maximum value and how you need to change your processes.”

He cautions that companies implementing AI should also be prepared to accept AI and automated processes will not necessarily deliver what a human-managed process will.

“You can’t assume the AI is going to do things exactly how you would do it manually. You have to expect a level of difference. It’s about understanding that difference and making changes until you are comfortable with the difference, rather than expecting it to be exactly the same,” he says.

“AI is really about looking at the efficiencies it can create within the business, whether that is speeding up a process, particularly a manual process or potentially removing manual steps, especially repetitive manual steps that are time consuming.”

Getting an early, quick win, is another key aspect, McPherson says.

“Being able to adopt an AI solution you can see works and delivers value will certainly give you confidence to look further afield into other AI solutions and to invest more in AI going forward,” he says.

Upskilling of teams is also a crucial factor, he adds, and something that can’t be done in isolation.

“With everything online now it is much easier for people to upskill and for employers to upskill their teams. But the courses need to be taken within the context of then applying what is learned straight away, within the business in the context of a pilot implementation,” he says.

“We are looking to automate as much as we can across our business, and AI is just one type of advanced automation, and it enables a degree of automation that wasn’t really possible before.”

McPherson will be speaking at the Digital First 2023 event on 7 March 2023 in Auckland. See more, and book tickets, here.