Generative AI is a revolutionary technology whose impact on society, industry and individuals is likely to be far larger than the invention of the internet or even the printing press and electricity. When considered together with developments in computing power, robotics and biotechnology, humans may be developing the technologies that replace much more than the work that we do. With a cultural and social technology able to take advantage of all the knowledge humans have accumulated and replace many of the tasks we train students to perform, what is the purpose and future of the university? With the coming wave of amplifying technologies in the hands of corporations more powerful and agile than legislators, what does leadership for the common good look like?
The timeline at the bottom of this post summarises the parallel developments in AI technology and the University’s approach since the release of GPT 3.5 stunned the world and its own developers in late 2022. With the disruption for education and the careers of our students clear in media, industry and governments reports from that time, the University has led the way nationally and globally in considering and now enacting the radical changes needed to maintain the integrity and relevance of higher education. Our sector-leading approach is reflected in the numerous invitations to present and contribute to institutions and government enquiries across the world, positive media coverage and through others’ adoption of our policies and innovative technologies.
This article was originally provided to the University’s Senate ahead of a student panel on the ensuring the integrity of our relevance of our degrees.
Lane 1: integrity
Assuring the community and employers that each of our graduates has the skills, attributes and knowledge that their transcripts claim is essential to our licence to operate. This assurance of learning is, of course, a legislative requirement through the Higher Education Standards Framework and central to all professional accreditation standards. More than this though, if students are awarded degrees based on passing assessments that a machine completed, we put the community and the economy in danger. If our credentials mean nothing, we have no licence or reason to operate.
Modern generative AI tools and interfaces such as undetectable browser plugins, tiny earpieces and cameras hidden in clothing mean that as much as 85% of our current assessment can be at least satisfactorily completed by a moderately tech-savvy novice. As demonstrated in the series of videos we made as part of a commission for TEQSA in 2024 (and already outdated), generative AI is able to accurately create content including written work, images, music, video and the spoken word and answer cognitively complex questions.
Secure assessment
The new Sydney Assessment Framework is now part of our integrity and assessment policies and procedures and is operating across all coursework degrees. In it, a secure assessment is defined an in-person supervised assessment of learning used to validate learning. Only assessment types which AI use can be reliably and effectively controlled through supervision are permitted. Whilst these include traditional modes such as in-person written and oral exams, they also include more authentic approaches such as interactive oral conversations, placements and in-person practical work.
Assessment plans
The power of generative AI and the miniaturisation of connected devices such as wearables (ear pieces, smart glasses and watches) means that supervision is becoming increasing difficult, expensive and intrusive. It is important that secure assessments are only used where they are required to assure course rather than unit learning outcomes. Faculties are currently completing assessment plans to map each pathway of each course:
- These plans are to be completed by the end of 2025 to map out where there is not sufficient secure assessment for each course learning outcome.
- By the end of 2026, the assessment and curriculum reforms identified need to be in place.
Ensuring integrity also means explicitly recognising the assessments and environments which cannot plausibly be secured. This includes all take-home and online assessments. It also means that relying on an honour code, statements that AI cannot be used or detection software is not good enough – they do not achieve security and serve only to reduce trust and agency. The impact on fully online courses should be noted – reflecting the unfortunate reality leading to the return of in-person job interviews by US companies. The number of sites offering help bypassing remote assessment security and cheating is growing.
Integrity for our education also means that we must ensure that our curriculum is future-ready and equips our students with the contemporary capabilities they need for the workplace. It also means that staff and student use of AI is transparent and ethical. Outside the secure assessments where AI is used by students, they must acknowledge its use. Where staff use AI tools to enhance their teaching, this should similarly acknowledge this.
Lane 2: relevance
AI is already affecting the job market, with recent reports of substantial reductions particularly for entry level positions. Unlike previous technology revolutions, the impact is expected to be mostly felt by white-collar workers, including university graduates and it is not clear at this time that there will be many meaningful new jobs created in its wake. With amplifying technologies such as robotics driving automation of everything from law to medicine and surgery, as many as 90% in Australia of roles could be affected by the time our current commencing students graduate. These disruptions could also deliver over a $110 billion productivity boost to Australia’s economy, according to a recent Productivity Commission’s estimate.
Open assessment
The other half of the Sydney Assessment Framework is therefore our open assessments. Constituting the majority of assessments, these are for and as learning. The effective and appropriate use of contemporary technologies is scaffolded for students as part of the learning process within units of study. Highly used assessment tasks such as online quizzes become opportunities for students to practice and apply their knowledge and gain feedback rather than to assure outcomes. The traditional written assignment maintains its role to develop communication and critical thinking skills using contemporary software. If these skills need to assured, this can only be achieved through a secure format such as an oral.
Future-ready curriculum
As generative AI and other technologies radically change how jobs are performed, our curriculum must keep pace too. Research and knowledge creation activities are no exception to this transformation. As part of the development of their assessment plans, disciplines have been asked to reconsider their course and program learning outcomes. In the longer term, the principles developed by the Curriculum Quality and Sustainability project must ensure the ongoing relevance of our portfolio of courses including lifelong learning opportunities for reskilling and, perhaps, occupying time freed up by automation.
Teaching practice
These assessment and curriculum reforms will require professional development for our educators. Generative AI also though can enhance the effectiveness of our teaching and enable our educators to spend less time producing resources and on administration. Our sector leading and award-winning Cogniti platform is being used by educators to design and implement “doubles” of themselves to provide students with 24/7 access to a range of agents including personalised tutorial support and actionable feedback on their written work. Cogniti also helps ensure us equitable access for students to modern AI engines. As educators themselves control how the AI answers students’ queries and the resources it uses, Cogniti also helps build trust and familiarity amongst our staff. Cogniti has won many awards and is now used by over 100 universities and over 200 schools globally.
Sector-wide leadership
The University has:
- Hosted two national roundtables with educational leaders to formulate responses to generative AI
- Run two international symposia on generative AI in education, attended by thousands
- Contributed to NSW and Federal government inquiries on generative AI
- Led an Association of Pacific Rim Universities whitepaper on approaches to generative AI in higher education
- Been recognised in Jobs and Skills Australia’s August 2025 report as a sector leader
- Shared its cogniti.ai generative AI platform with hundreds of educational institutions globally
- Received national and international awards for cogniti.ai, and formed a close partnership with Microsoft through the platform
We will continue bringing together educators and other voices to formulate ways ahead for education in the era of AI. Watch this space!
Timeline of global developments in generative AI and at the University
Date | University of Sydney | Global developments |
November 2022 |
|
|
December 2022 |
|
|
January 2023 |
|
|
February 2023 |
|
|
March 2023 |
|
|
April 2023 |
|
|
May 2023 |
|
|
July 2023 |
|
|
August 2023 |
|
|
September 2023 |
|
|
November 2023 |
|
|
December 2023 |
|
|
February 2024 |
|
|
March 2024 |
|
|
May 2024 |
|
|
June 2024 |
|
|
July 2024 |
|
|
September 2024 |
|
|
October 2024 |
|
|
November 2024 |
|
|
December 2024 |
|
|
January 2025 |
|
|
February 2025 |
|
|
May 2025 | · Anthropic releases Claude 4 | |
June 2025 |
|
|
July 2025 |
|
|
August 2025 |
|
|
September 2025 |
|
|
December 2025 |
|
|
December 2026 |
|
|
Now |
|
|
Then |
|