UNE’s new AI-focused unit aims to reshape the way we work

The arrival of generative AI for the masses has, in the space of a few years, overturned old assumptions about the process of education.

Rather than circle the wagons, UNE has adopted a policy of meeting AI head-on, with the aim of putting the technology into harness in service of the University.

In 2024, UNE added a new unit, LabNext70 (LabN70), which has been charged by the Vice-Chancellor with introducing practical AI applications into the University workplace.

Specifically, LabN70 is exploring how AI can be used to improve the student experience and lighten the load of drudge work on staff. Sometimes that will involve addressing a well-understood problem with conventional AI tools, but the Lab also has a brief to think outside the box and build AI-centric tools to improve the learning experience in new and novel ways.

LabN70 is led by Associate Professor Aaron Driver, who has a background in communications, marketing, tech and academia. He will draw on these skills to support the introduction of a technology that carries considerable uncertainty along with its promise.

“Our goal is to help everyone else build AI into everything, everywhere it can add value, in a way that is humanising for the people who use it,” Aaron says.

“We spent a lot of 2024 building the foundations we need to work from and undertaking some tentpole projects. In 2025, the Lab’s doors will be visibly open and staff can start bringing their problems and solutions to us.”

Aaron wants staff to think about how AI might be enlisted to undertake necessary but time-consuming process work.

As an assistant, AI’s lack of personhood can be an asset. It doesn’t get miffed at having to repeat something, watch the clock, or carry personal baggage. It also scales with infinite flexibility, so it can be used to take care of a small personal work burden, or build a whole-of-institution solution.

This year, LabN70 will be hiring ‘prompt engineers’ – experts who understand how to interrogate the Large Language Models (LLMs) that AI is built on, in ways that produce an expected result. (To the uninitiated, AI is capable of producing any number of unexpected and unsuable results.)

Staff wanting to bring AI into their workplace will be able to submit projects via the Lab’s pipeline to receive dedicated support and/or help building an AI assistant. Often, an AI assistant will be useful in more than one context, so that over time LabN70 will build a standing workforce of AI applications that can be deployed for any work better suited to digital intelligence than human.

LabN70 has also undertaken several larger initiatives that are already bearing fruit in the areas of teaching and learning and student recruitment.

A first step in embedding AI throughout the UNE workplace has been to adopt the Cogniti AI platform, which will be installed on UNE’s Azure system early this year.

While anyone can pay to use commercial LLMs, these platforms do not have guardrails to protect staff and students from misinformation or AI hallucination. Cogniti enables academics to access the power of AI, with guardrails.

A University of Sydney startup, Cogniti is “built by educators, for educators”. It provides a number of agents that can take on various roles, like that of a tutor, and help students access information from a contained selection of material without requiring input from a lecturer.

Aaron says early results have been encouraging. “The academics who have adopted Cogniti are getting dozens of students making hundreds of interactions with the AI, to a hugely positive response.”

As Cogniti is built specifically for the teaching and learning environment, Aaron has also negotiated a “fit for purpose” AI platform that can service the needs of professional staff. Contracts have been signed, and an early pilot of this new platform will begin this month.

LabN70 has also recently launched a trial of an AI agent it calls Belong, aimed at addressing the ‘enrolment melt’ of prospective students.

Belong was built for a particular challenge. When UNE sends out letters offering a place to prospective students, about half of those who accept the offer never go on to enrol.

“It’s a high loss rate,” Aaron says. “So Project Belong involves messaging those lost students and inviting them to come and chat with our AI agent.”

“The agent is trained to support student self-efficacy, which we know is a big factor in those dropping out – especially in the equity student cohorts we are targeting.”

“Students need to believe they can take this big scary step into university. The AI helps them understand how to enrol, how many units they should take, how they can structure their study. It’s helping them overcome those barriers of self-belief.”

Logs of subsequent interactions with the AI are manually checked for veracity. So far, Aaron says, 99% of the time the AI is offering accurate advice.

A third project, now closed, is a student competition with the objective of identifying “AI-savvy” students at regional secondary schools and universities. “We want to get them involved in the lab as interns, and start getting them engaged in this business of transforming UNE for the next era of education. Winners will be announced soon.”

Mass adoption of generative AI has only been possible for about two years, and LabN70 has existed for less than half of one. The technology has already wrought profound changes on the way we work, Aaron observes, but the biggest challenges are before us.

“The growing capability of LLMs, and the application layers wrapped around them, will be important, but what’s most important to the future of AI is how we shape it through usage. If we ignore it at UNE, or treat it as a sideshow, then others in the sector will do the shaping. We’d rather we did it our way, and build humanising, empowering solutions to serve the region and secure the future of our university.”

Image: Unsplash