Business needs to get with the AI program
As artificial intelligence becomes increasingly user-friendly, the benefits of adoption are clear
-
-
Across Australia’s private and public sector, organisations are using Artificial Intelligence (AI) to swiftly analyse vast volumes of data to uncover insights, find solutions to complex problems and build services that were previously impossible.
As experience with AI-based technologies grows, businesses are also learning the skills needed to apply them most effectively and overcome the challenges they pose.
According to IBM’s director for data science and artificial intelligence, Ross Farrelly, even though the origins of working AI stretch back to breaking codes in World War II, three concurrent factors have thrust it into prominence.
“First, there’s a huge explosion in the amount of data that’s available based on the digitisation of many aspects of our society,” Farrelly says. “Then we’ve got the compute power available through the cloud, meaning that many companies, even small- and medium-sized organisations, now have access to high compute capabilities.
“And third, we have the sophistication of the algorithms. Many of those algorithms had been previously developed, but they were too complex to run at scale. But again, with cloud computing, that problem is largely solved.”
The convergence of these three factors is enabling more organisations to solve difficult problems using AI. One example is a large Australian telecommunications firm, which had adopted AI technology from IBM to assist with a range of functions including how it engages with customers.
The organisation wanted to deliver a faster response to customers from the 50 million interactions they engaged in each year, as even a 5 per cent increase in customer retention could increase business profitability by 75 per cent.
The strategy was to automate simple interactions, reduce the internal transfer rates, and provide a better experience to customers. The first application of AI in this capacity was a virtual assistant or “bot”, built using IBM’s Watson AI technology, which addresses customer queries by interpreting their needs and then routing them to an answer or a human agent.
The success of the chatbot led the company to examine how AI could be deployed in other areas of the business, including the creation of a bot to help field service engineers perform troubleshooting. Not only did this reduce costs, but it also provided better outcomes across the business.
Farrelly said the interface between humans and systems was proving to be a popular area for investment in AI technology across many organisations.
“Machines are now becoming a lot more capable of understanding and responding in natural, straightforward, plain language,” Farrelly says. “And that means that they can be used for things that they could never be previously used for.”
IBM was looking to take this concept one step further through Project Debater, whose goal was to train a computer to debate with a top-level world-class debater on any topic.
“If you think about how complex that problem is and the hundreds, if not thousands, of different sub-problems that have to be solved in order to hold a debate, it’s quite mind blowing,” Farrelly says.
Projects such as this could have very practical real-world applications. Farrelly says IBM was already working with a US partner that had developed a system to help with the early stage of drafting of legal documents. He adds that there were many examples of AI proving adept at performing menial tasks that humans find tedious.
“A lot of that is very repetitive, pretty mundane and not particularly engaging for a lawyer,” Farrelly explains. “That can be done by artificial intelligence.”
IBM’s AI technology was also playing a key role at Australia’s Woodside Energy, where CEO Peter Coleman hopes AI can slash Woodside’s $1bn annual maintenance bill by 30 per cent, while also helping the company better manage logistics at its operating sites.
He has also expressed his desire to see AI play a key role in Woodside’s plans to create “the plant of the future” where AI could provide timely intelligence and insights to empower the workforce and optimise decision making.
But while AI has proven adept at providing answers to complicated questions in a timely fashion, Farrelly says it was important to ensure AI systems were not inadvertently incorporating bias into their results. This concern had led to a rise in interest in the concept of explainable AI, especially in situations where the results of AI queries ran counter to established wisdom.
“When that happens, then the output from the model will be challenged quite vigorously,” Farrelly says. “And that’s where explainability really comes in.”
As a result, he says IBM was making explainability a key part of its AI offerings,
to demystify the results that systems were generating. But explainability of the outputs of AI is just one challenge to be overcome in the accelerating adoption of AI.
According to Jason Leonard, IBM Australia’s partner in cognitive computing and analytics, it is essential that AI projects include a blend of business skills with data science skills, to ensure AI is being applied to tasks that offer real value to the business. He says some organisations struggled to get out of the experimental phase of AI implementations.
“The model also needs to be explainable in ways that the line-of-business understands.” Leonard adds. “Some do a wonderful technology proof-of-concept, and maybe get it into production and it works well, but if you ask them what is the impact to the bottom line or the employee or customer experience, they may not be able to answer.”
Those who were successful tended to be the ones who understood upfront what they were trying to achieve, tracked results along the way, and created a road map towards a broader vision.
“Then they are able to get repeat investments going as well,” Leonard says.
This approach could be seen working well at Woodside, which had brainstormed the possibilities for AI, and started with a single project that documented lessons learnt on projects. This was then extended into occupational safety, and then on to HR.
“The data might be different, but the overall approach was the same,” Leonard says. “To help with that line of business connection they went from talking about the technology to innovation forums to incubate ideas and prototype them with a business context in mind so they could be scaled out.”
By bringing together both business users and data scientists and aligning them to a common cause, Leonard says amazing results could be achieved.
“What excites me is the creativity involved — you never know what is going to pop up next,” Leonard says. “This is a great opportunity for growth, and it is going to change the ways business is done … it is just a matter of which organisations are most excited about that.”
-
Read The Australain’s policy on commercial content here.