By Professor Daswin De Silva, Deputy Director, Center for Data Analytics and Cognition
Artificial intelligence (AI) heralds a new era in workforce productivity, revenue growth and profitability across all industrial sectors. Generative AI continues to lead this transformation, building exponentially on its foray into the public domain with the release of ChatGPT in late 2022.
ChatGPT was an overnight “success” by disrupting several areas of work, including higher education, media, creative arts and healthcare. Generative AI is commonly defined by distinguishing it from what we now call conventional AI; its ability to “generate new, non-trivial, human-like, accurate, and seemingly meaningful content,” this content has its own acronym “AI Generated Content (AIGC).” On the other hand, conventional AI is task-oriented and well-defined for what can be grouped into problems like prediction, classification, association and optimization.
The rise of generative AI is gradually being recognized as a general-purpose technology due to workplace innovation and market spillovers. Most knowledge works are heavily exposed to generative AI, with About 80 percent of the workforce has at least 10 percent of their tasks automated and nearly 19 percent of occupations have a 50 percent higher risk of exposure.. Generative AI is highly competent at specific tasks, meaning that overall work activity can be augmented instead of automated by segmenting it into tasks collaboratively assigned to AI agents and human workers.
As the hype builds, governments and organizations have strategized approaches, services, training and recruitment to take advantage of AI opportunities. Recognizing AI as a technology of national interest, the Australian Government has invested in several initiatives such as the National AI Centre, the AI Graduate Program and the AI Adoption Centres, as well as adopting an approach based on risks which is similar to European AI law for ethical and responsible AI.
However, when it comes to industry engagement, these initiatives are largely centralized, despite AI’s fundamentally decentralized growth potential (so AI is not a universal technology). While welcoming and supporting these first steps, it is equally important to understand and analyze the missing elements.
Although commercial variants of generative AI are recent, AI research, technology development, education and training has been the mainstay of Australian university research centers and laboratories for many decades. Closer to home, La Trobe’s experience in AI spans teaching, supervision, publications, grants, as well as deep engagement in the industry that delivers results impactful, like an Australian first. Energy AI Platform for Net-Zero Carbon Emissions, Bespoke AI Micro-Certificate Training for Optus Employees, an inclusive approach to the AI lifecycle and many more.
AI also has a trust issue in the public eye, which half can address through open and transparent means.
Unlike the US and Europe, Australia does not have an active, risk-aware startup culture capable of investing in and supporting AI innovators. This has resulted in little intersection between academics whose ideas cross paths with industries seeking innovation. Most often, technology consultants are preferred over universities because they take the time to understand and translate an industrial need into a technical solution delivered on time. The academic AI community will benefit from training and mentoring to look beyond just the technical performance of an AI model and to the practical value of a working AI system. A similar approach should be taken for AI training, where additional efforts should be made to develop tailored AI education programs that seamlessly integrate with the workplace, technological equipment and to workforce skills. A notable example here is the gradual disappearance of the need for programming skills and the emerging need for rapid engineering when creating and using generative AI.
From an industry perspective, the difficult early stages of AI require a leadership mindset that embraces “digital, data, analytics and AI.” A working digital infrastructure with a centralized repository of all data is a typical starting point for most organizations considering the move to AI. This data repository manages analytics dashboards, formal reports, ad hoc queries, and serves as an enabling layer for conventional AI capabilities of prediction, forecasting, etc. For example, AI models for revenue forecasting require large volumes of high-quality training data of past transactions that must be reliably stored in the all-data repository and calculated in the next layer, l digital infrastructure. In this linear approach, mature expertise in conventional AI will gradually inform the transition to Generative AI. An alternative to linear is moving “AI first” to generative AI. For example, starting with “Gen AI hackathons” where all employees receive initial training and short-term paid access to Gen AI subscriptions and assistants to innovate their own workflows and work activities against metrics. basic performance. The first applications of this approach would be in content creation, meeting minutes, targeted marketing and project management.
Just two indicators are enough to understand the transformative potential of AI, the number of new AI models released and the number of new applications of existing AI models. Both indicators follow exponential trajectories. To compete with thriving AI economies elsewhere, all Australian industry sectors must take rapid action to synergize with the “better half”, the academic AI communities, to initiate, inform, to support and enable AI innovations for industry progress.
La Trobe Industry Communications and Media Inquiries: industrial.engagement@latrobe.edu.au
