At Neo4J GraphSummit in Sydney, Tim Sheedy, Vice President of Research and Chief AI Advisor at EcosyStm, shared insights into the adoption status of Generated AI (Gen AI) in Australia and New Zealand, highlighting both the possibilities and challenges the region faces.
“The majority of organizations are somewhere in the middle,” Sheedy admitted to the audience. “Some organizations are deployed within business units. Some organizations have come out across the organization. But to inform you of most of them, they are not AI-first companies.”
Sheedy said that while most companies are still in the “integration phase” and focus on setting up the data pipeline and AI infrastructure needed for transformation, they are focusing on the fact that there is no change in AI-driven organizations yet.
“76% of companies say they're consolidating. This means they prepare data, develop AI strategies, pilot AI applications, but don't think they're fully converted with AI.”
Despite these challenges, Sheedy highlighted several successful use cases of AI adoption, particularly in the customer experience.
“Organisations that did well reported a 52% improvement in customer experience scores,” he said. However, he emphasized that AI is primarily used to improve operational efficiency, and that companies are looking for ways to do more in less ways.
“Business leaders want to do more with less or better with today's stuff. They all see AI as a great opportunity to drive productivity.”
Sheedy pointed to an increasing trend in AI applications in operational tasks such as intelligent document processing, inventory management, and code generation.
“More than half of the organizations already use Gen AI for these tasks, and about 60% use it at their help desk,” he explained. However, he revealed that AI has not yet spread to all functions within most companies.
“They aren't saying that all of the document processing is wise across the organization, but they use it somewhere in their business.”
Looking ahead, Sheedy observed the rapid growth of AI use.
“The use of AI is on the rise and is growing rapidly,” he said. “In addition, areas such as cloud resource allocation, software development, fraud detection are set up to experience more AI interventions.” He also shared examples of multinational companies using Gen AI to significantly reduce the time to change proposals, from six weeks to just two days. “By using Gen AI, they're significantly reducing that time to put together the first draft of that time,” Sheedy added.
However, Sheedy hastily warned of the risks of AI adoption.
“72% of consumers avoid brands after AI errors. I think this is the same as employees in your organization,” he warned.
“You've had a bad experience with a chatbot. You probably won't use that chatbot again in six months.
He also referenced the infamous Zillow case. There, AI-driven real estate algorithms lost USD 500 million due to flawed forecasts.
“If AI doesn't work, AI can have serious economic consequences,” Sheedy said.
Regulatory and organizational challenges are also delaying adoption. Sheedy stressed that while Australia's regulators are in more consultation, the cost of making things go wrong is important. “It's really important to get Gen AI right for the first time so that you don't lose that trust, you don't lose that money, you don't lose that time,” he added.
Skill shortages and data challenges are the main barriers to AI adoption in ANZ.
“Skill shortages and data challenges are much more sustainable and difficult to deal with,” Sheedy explained. “At this point, there are thousands of organizations in Australia trying to hire people with non-existent skills, or if they exist, they aren't trying to get them for what they're paying for.”
Sheedy also highlighted the need for better data management. “Our data is not clean enough. The data doesn't understand the context,” he said. He advocated a new approach to data, particularly through the use of graph databases, which provide a “semantic AI” layer that allows companies to understand the relationships between data points.
“Graph databases provide the same functionality as data that places data within an organization within these language models. This helps organizations understand the relationships between data.
Sheedy also pointed out that adoption of AI is often hampered by the difficulty of modifying core business processes.
“We need to change the processes within our organization to make AI do better or different,” he said. This challenge is particularly evident when an organization needs to integrate AI with legacy systems.
At the organizational level, Sheedy emphasized the need to better understand AI.
“The people in our organization don't understand our company. They don't understand what AI can and can't. We really need to be skilled widely throughout our business,” he said.
Drawn from the cloud computing experience, he suggested that AI adoption only accelerates if an organization trains the entire workforce on technology capabilities and limitations.
“I believe that AI will be hyper-growth when AI starts with the board and CEO and starts training all employees across the business on what an organization can and cannot do with the business.”
Sheedy also emphasized the importance of embracing new technologies like graph databases to stay competitive.
“Graph DB is everywhere within our organization,” he said. “We're at the beginning of a massive wave. It reconstructs how we understand the web and data. We need to get early on and drive conversion.”
