Wait for the Aussie AI Adoption Boom
- By Lachlan Colquhoun
- December 04, 2023
Australian corporations are preparing themselves for the tsunami of change coming with implementing AI at scale, but several challenges, ranging from regulatory uncertainty to data ownership and management, are a more immediate focus.
This was one of the consensus themes from a recent Sydney roundtable hosted by real-time generative AI solutions provider DataStax in collaboration with CDOTrends.
Invited guests from the financial, investment, and media sectors joined DataStax subject matter experts in a wide-ranging discussion on how AI was impacting their organizations and discussed how far along they are in the journey from ideation to impact.
Early times
Warren Schilpzand, the area vice president for Australia and New Zealand at DataStax, described the current AI landscape and set the scene for discussion, outlining how Google's Bard and Chat GPT had popularized AI, driving "nearly every board" to ask their technology leadership to respond with an AI strategy.
Schilpzand likened this to the beginning of the cloud computing era that saw many allocating significant investment and resources to solve some problems but remained disappointed in value. He asked roundtable participants how AI would have helped.
Daniel Entin, the head of customer engineering for digital natives at DataStax partner Google, emphasized that strong data foundations were the foundation of AI success.
“80 to 90% of the work is with data,” Entin said, a point to which Schilpzand returned later in the discussion.
There was also a discussion about AI definitions and how AI was split into generative and predictive AI.
Predictive AI uses machine learning models to train for inference, while generative AI entails feeding a large language model (LLM) with a set of prompts, with the output presented back to the user in natural language.
Schilpzand said that as they prepared for the era of AI, organizations should "invest in their data future."
"The tsunami of AI is coming, and if we are going to be ready, we need to have our underlying platforms ready," he said.
"AI means data and huge amounts of it, and to make use of this technology, you need to be able to access it and access it the right way."
Regulating outcomes
Regulation was a key topic of discussion, particularly in financial services, and its impact on innovation in the space.
The current approach of regulators in other global jurisdictions, such as the U.K., has created a "sandpit" where technology can be tested, and lessons learned. Attendees showed enthusiasm for a similar approach in Australia.
Attendees shared enthusiasm for regulation to be outcome-based rather than internal-facing, which drove a discussion on the difference between established legacy organizations and digital native startups. Some participants believed startups could move faster and innovate with agility on AI "while other companies are moving at our old speed.”
For legacy organizations, there can be established revenue to protect, impacting a team's ability to get an AI project up and running.
Startups, in comparison, could have little to lose in this regard.
One advantage legacy organizations had, however, was that, in general, they had large volumes of data that would be used for AI projects, even if they were also limited by their technology stacks and access to talent.
Once again, however, there were regulatory and data ownership challenges to negotiate.
Some participants said they were "tinkering on the edges" of what AI could do, such as looking at "smarter ways" customers could respond to a product design specification (PDS) document.
Most were looking forward to a time when AI could enable them to "do tomorrow what we can't do today" and find new insights unavailable to humans, largely in customer-facing applications that were predictive for the organization and delivered tailored offers and suggestions to customers.
RAG rises
For many organizations, the impact of AI would be "qualitative but incremental," according to one participant, who said the revolution would not come from the technology itself but from a "step change in the way the business operates."
This was part of a long journey to scale new use cases, although DataStax's Schilpzand said that the firm had worked with companies in the U.S. to "turn around generative AI cases in a matter of hours" from already available data.
This was through the use of retrieval augmented generation (RAG), leveraging the capabilities of an LLM with a data retrieval system that provided non-public or proprietary data to a model in a protected fashion.
RAG is a hybrid framework that integrates retrieval and generative models to produce relevant and usable content, such as recommendation engines using AI in real time.
For example, this has also been used in building a search capability for all the documents in a law library. In cases where all that was required was for the LLM to understand a question and generate a real-time response.
This was presented as a quick way to derive outcomes and benefits from available data.
Australia's pattern of technology adoption suggests that while there might be an initial lag on implementation, once proven, AI momentum could pick up rapidly; participants agree that this was likely to be the future pattern with AI.
To download in PDF, click here.
Image credit: iStockphoto/Deagreez
Lachlan Colquhoun
Lachlan Colquhoun is the Australia and New Zealand correspondent for CDOTrends and the NextGenConnectivity editor. He remains fascinated with how businesses reinvent themselves through digital technology to solve existing issues and change their business models.