How Gen AI can be Australia’s productivity accelerator

How Gen AI can be Australia’s productivity accelerator

By Adam Beavis, Vice President & Country Manager, Databricks


Labour productivity growth has slowed in Australia since the mid‑2000s. In fact, records by the Australian Treasury show that in the decade to 2020, productivity growth was slower than it had been in the 60 years prior.

Australia’s new productivity commissioner Danielle Wood is optimistic that Artificial Intelligence (AI) will be the answer to dwindling productivity. At the Australian Securities and Investment Commission’s annual summit in November 2023, Wood said that AI could help repair Australia’s lagging innovation record, unleash waves of productivity gains and lift the output of workers across various domains. By her estimate, AI could affect as much as 80 percent of the economy.

But before businesses can leverage AI they must overcome a variety of challenges including costs, data security, and data disparity. Luckily, the democratisation of AI is helping to curtail these issues so that any business, from the largest enterprises to the smallest mum-and-dad operations, can reap the benefits that AI can offer such as spur fiscal growth, foster innovation in the workplace, improve customer and employee experiences, without significant threats to jobs.

Challenges in adopting AI to unlock productivity

Despite business leaders’ enthusiasm around enhancing productivity with AI, the adoption process can come with some challenges. In many cases, the costs and human resourcing needed to implement AI alone are enough to put this transformative technology, and its benefits, out of reach — especially for small businesses.

This is because building AI software from scratch requires expertise from technologists, data scientists, and engineers who know how to code – insights which can be hard to ascertain by smaller enterprises. Furthermore, AI models also require considerable regulation, including risk assessment and mitigation, clear documentation around data sources, bias, and so on. As these standards evolve, it will become more challenging for organisations of all sizes, but especially smaller ones, to keep up and ensure compliance. This can act as a major barrier to AI implementation.

While business leaders may be tempted to use publicly available AI models to gain the benefits without the costs, doing so comes with risks. For example, any private data shared with the public AI provider may be stored online and used for further learning, increasing the possibility that their proprietary data could be leaked. We’ve seen instances where companies have experienced accidental leaks,  like when a Samsung engineer inadvertently uploaded sensitive data and source code when using the ChatGPT platform.

A further challenge lies in collating the extensive amounts of data required to train an AI model. In many cases the data businesses have available to train AI models is a combination of structured and unstructured data —which is often spread across various disparate systems. The lack of uniformity of data types and presence of data silos make it difficult for companies to fully exploit the benefits of their AI models.

How democratising AI removes barriers to adoption

Open source AI platforms play a pivotal role in breaking down barriers to AI adoption for businesses of all sizes across all industries. With open source AI software, a lot of the hard work is already done, and organisations can simply build their AI models on top of them. Generative AI also enables the democratisation of the technology allowing any organisation to create AI use cases by prompting the systems what to do — without coding or technical expertise required.

Leveraging open-source technology to build in-house AI models also removes the risks associated with using third-party AI services and helps organisations protect their sensitive data and IP. Given that the in-house AI model can be trained on company-specific data, without the risks of leaking sensitive information, there will be greater accuracy in results and there can be a more nuanced way of playing with the data.

Using open source AI can also help to eliminate the silos that complicate data and inhibit AI functionality. Software systems like Databricks’ Data Intelligence Platform unify data from various different formats ensuring full visibility and fine-grained control throughout the AI workflow. Furthermore, simplifying your data lakes through combining them can enable organisations of all sizes to overcome certain barriers to AI implementation as the process standardises the entire machine learning lifecycle. This allows core teams to mine, prepare and govern data and empower them to implement and scale it where needed.

Examples of AI at work

Several Australian businesses have already begun implementing AI, from streamlining processes to enhance customer experience to implementing task automation and reducing data centre call times to improve productivity.

NAB is devising and implementing AI strategies to make the process of loan approvals more efficient, thereby freeing up its frontline workers and enhancing customer satisfaction. The company’s HR department is also leveraging the technology to help create new training courses and automating certain processes, effectively cutting the time it would normally take to complete the task down from weeks to days.

Suncorp also provides another compelling example of leveraging AI to improve productivity and helping its team respond faster to natural disaster claims by combining geospatial data, customer policies and weather information to allow the company to identify and outreach to customers who may be impacted by a weather event ahead of time.

Leveraging geospatial data and AI has also helped Suncorp to remove 50% of questions from its quotes process. This has reduced call time in data centres by 91 seconds, ultimately improving the overall customer experience and reducing operational costs.

Final thought

As Australia grapples with slow productivity growth, the transformative potential of open source AI has emerged as a beacon of hope. By providing readily available AI software, open source platforms empower businesses of all sizes to harness the power of AI without the costs, security risks and data disparity issues that are usually associated with implementation. In short, the democratisation of AI will help drive productivity and expedite innovation in Australia for many years to come.