AI is risky but we can’t ban it say Go8 universities

Educators must find a way of working with generative artificial intelligence in spite of the risks, universities say.

We can’t ban generative AI in universities, says the Group of Eight.

Australia’s top research universities have told a parliamentary committee that educators must find a way to meet the challenge of working with artificial intelligence and that a ban on its use is not feasible or desirable.

But other submissions to the inquiry into generative AI in education by the House of Representatives Standing Committee on Employment, Education and Training point to the difficulties universities have in dealing with AI, particularly in maintaining integrity in assessment.

Group of Eight chief executive Vicki Thomson, who represents the main research universities, will tell a committee hearing on Wednesday that generative AI poses opportunities and risks.

Risks include cheating, privacy issues, intellectual property violations, introduction of bias and lack of equity, while there are opportunities to improve teaching and make research more productive.

“Our view is that prohibiting the use of generative AI in higher education is both impractical and undesirable and these principles will help us take advantage of AI whilst ensuring academic integrity and ethical use of generative AI,” Ms Thomson will say.

In its submission to the inquiry Universities Australia said greater regulation of AI in education “may be appropriate”. It warned of the risks of over-reliance on AI including that data produced using AI tools “can still be far from accurate”.

Deakin University academic Phillip Dawson told the inquiry in his submission that AI detection software, which is being used by some universities, is not currently effective.

“A review of 14 tools for detecting generative AI in student work conducted by a consortium of researchers from eight countries has found that none of these tools function well enough for use in education,” said Professor Dawson, who is an expert in assessment and digital learning.

He told the committee AI detection tools should not be used in education until there was independent evidence that they worked.

In its submission UniSQ listed many benefits for students from using generative AI including the ability to offer a “highly personalised learning environment” and automated “tutor-like” support for students with on-demand feedback.

UniSQ also said AI could be used by staff to do administrative tasks, allowing them to concentrate on higher-value activities. AI could also assist academics in assessing students’ work and in creating learning content.

But it acknowledged several risks, including students using AI to do assignments which “requires assessment to be re-imagined to ensure it remains valid and verified”.

Monash University told the inquiry in its submission there were risks to both overusing and underusing AI in higher education. If used as a shortcut in learning or assessment tasks there was a risk to academic integrity, the university said.

However, underusing AI was also a risk, it said.