If we didn’t know this already, this week’s launch of China’s revolutionary DeepSeek v3 chatbot shows that AI will continue to change and disrupt expectations in 2025. But beyond getting faster, cheaper chatbots, for me the critical question is this: Will 2025 be the year that AI moves from “potential” to “productivity”? Or asked in a different way, where is Generative AI currently placed on the famed technology ‘hype cycle’? Are we entering the “Trough of Disillusionment” – or are we instead moving into the “Slope of Enlightenment”?
In Australian and New Zealand higher education, these questions are highly relevant and pressing for decision-makers. As one expert that I spoke to recently said, “Higher education is one of the last industries to experience the full force of digital disruption – AI will change that within the next 5 years.”
Where is AI having the biggest impact for higher education?
Over the past two weeks, I’ve spoken to several Australian higher education pioneers about where they are already investing in AI, and where they see future potential. There were a few really interesting conclusions:
AI productivity tools for staff are delivering “broad but shallow benefits”. Adoption is patchy, however, and performance benchmarks are not clear – these tools are particularly beneficial in three key departments: IT, Marketing and HR. This aligns with private sector trends, with these three disciplines being the top 3 areas for AI adoption globally . However, adoption is slow and inconsistent, and there’s an opportunity for institutions to more actively direct and train staff to adopt these technologies and track their performance in doing so.
AI has massive potential as a tool to deliver personalised learning experiences -- but true transformation is a way off. One the biggest challenges that Australian higher education faces is how to deliver personalised experiences to students, especially in the largest subjects. AI could fill this gap, in the same way that it’s doing in health care as a way to offer personalised diagnostics and care. Australian and New Zealand universities have been slow to act on this, compared to global leaders such as Arizona State University.
AI promises a step-change in administrative and operational efficiency, but data quality is a critical barrier to success – university decision-makers see huge potential for AI to improve and automate key administrative processes, particularly in areas such as Student Administration. Different types of AI are relevant here beyond Generative AI, particularly Machine Learning, Predictive Analytics and even Natural Language Processing. However investments in data quality are critical to realising benefits - in a recent survey of UK business leaders, only 27% of leaders agreed with the statement that “my company’s data is of sufficient quality for use in training or tuning AI models”. The universities that invest in addressing this challenge will reap rewards fastest.
Some applications of AI are questionable, but have potential if key issues are addressed:
Student support and wellbeing – the use of AI as a tool to provide student support is increasingly popular, and there are some good examples (here and here) of AI offering advantages like 24/7 support, and enabling staff effort to be more targeted. However at the same time, student feedback has indicated a level of scepticism about the usefulness of AI tools for monitoring their wellbeing and provide recommendations.
Assessment and marking – opinions are divided on the use of AI for assessment and marking. Universities have long seen this as a potential application of AI, and experts we spoke to see AI as a way to improve consistency in assessment. They asked the question “does a human being mark the 300th essay the same way that they mark the 1st essay?” However, others perceive ethical concerns and some technical challenges with using AI in this context.
Curriculum development – students have also expressed scepticism on the efficacy of AI-generated content. One of the biggest concerns about AI more broadly is the coming tidal-wave of low-quality, AI-generated content, and this applies within higher education as much, if not more so.
How can decision-makers use this information to guide AI investments and strategic planning?
Given the above observations, how should higher education decision-makers approach AI strategy and investment planning? Drawing on our own client work, conversations with experts, and research into global best practices, we have developed a matrix of Benefit Potential and Solution Maturity which maps AI use cases for higher education into one of 4 categories:
“Stars” - Scale up and drive adoption – these use cases are High Benefit, High Maturity, and institutions should invest to scale up and drive adoption across staff.
“Diamonds in the Rough” - Invest to experiment and learn – High Benefit, Low Maturity, these use cases should be a focus for institutions to rapidly test and iterate to see if they can be matured into more scaleable solutions.
“Slow but Steady” - Pay careful attention to quality and cost – Low Benefit, High Maturity, these use cases are reliable but benefits are incremental rather that transformative. Only invest if the solution is cost effective.
“Possible Sinkholes” - Avoid for now but continue to monitor – Low Benefit, Low Maturity, these use cases are not delivering clear value at the moment, but this could change.
In the chart below, we map 8 broad use cases of AI in higher education onto this matrix. These are relatively broad categories and we hope to add more detail over time, but even at a high level this offers a useful starting point for decision-making and prioritisation.
Figure 1 - AptoNow’s Benefit/Maturity Matrix for AI in Higher Education

So … will 2025 be a tipping point?
With some exceptions, Australian higher education is notoriously conservative when it comes to adopting new technologies, and this also applies to AI. And not without reason – in many cases the AI hype exceeds the reality. For example, Gartner has predicted that at least 30% of Generative AI projects will be abandoned after proof of concept by the end of 2025.
But I'm going to nail my colours to the mast and say that 2025 is a critical time for realising the transformative potential of AI. If you believe, as I do, that AI will transform the future economy, it’s critical for Australian higher education to develop clear strategic priorities for AI, and invest early and often. Pioneers in this space (such as La Trobe University, who AptoNow is working with) are already moving ahead at pace -- and laggards risk being left behind. AptoNow works with leading Australian and global universities to develop AI and data analytics solutions to their biggest strategic and operational challenges. Please reach out to us directly at info@aptonow.com (or to the author directly at tom@aptonow.com) if you'd like to discuss how we can help your institution.
Comentarios