Research apprenticeships are a valuable part of Australia’s innovation, research, and development ecosystem. At their core, they involve experienced academic supervisors supporting novice researchers. But the nature of this apprenticeship is quickly changing, increasingly powered by AI technologies.
Such apprenticeships can include students enrolled in a Higher Degree by Research (HDR) or Doctor of Philosophy (PhD), as well as Early Career Researchers (ECRs). This training is designed to help next-generation researchers develop the knowledge, judgement and critical-thinking skills to navigate their future careers.
Starting with integrity
Associate Professor Janine Arantes from Victoria University is an HDR supervisor and expert on AI ethics, psychosocial risk and education policy.
When taking on new research apprentices, Arantes says conversations “absolutely have to start with integrity”. That includes consideration of ethical principles and adherence to professional research standards. Without attention to research integrity, a multi-year project like a PhD isn’t worth the paper it’s written on. Some experts believe it might even be an appropriate time to revive the discussion of viva voce examinations for PhDs in Australia.
As an HDR manager, Associate Professor Maggie Zhai from RMIT University believes supervisors must guide their students as technology shifts. As HDR manager in the School of Science, Zhai encourages supervisors to keep up with AI to best provide critical training and insights to the next generation of researchers.
Zhai believes that with supervisors taking a leading role in AI use and literacy, they can guide students on the things “feared most for research apprenticeship”. This includes developing judgment, thinking critically and avoiding plagiarism.
Many Australian HDR supervisors echo these sentiments, according to forthcoming research conducted by Jisi A, a Victoria University PhD candidate supervised by Arantes. “Supervisors see clear benefits in using generative AI for efficiency and research support, but they are also cautious about issues such as accuracy and research integrity,” he says.
Levelling the field?
There is clear tension over whether – and how – generative AI (genAI) might level the academic playing field. Zhai, Arantes and their students see benefits and risks, depending on the circumstances. They outline a range of scenarios.
GenAI can generate “really beautiful diagrams, schemes, data or figure plots,” says Zhai. This allows novice researchers to spend less time planning their communication, though it can lead to questions about the competence and capability of the student.
Unfortunately, research apprentices who have the most to gain from AI assistance may also be the least equipped to recognise when AI is wrong, according to Arantes. Her approach is to encourage necessary friction. “You have to be exposed to ideas you don’t necessarily agree with,” she says. This helps to develop critical thought and rigorous research.
Dr Lorenn Ruster, an ECR at the Australian National University, wonders if genAI might level the field for researchers in another way. She recalls how the scaffolded support of the PhD research apprenticeship helped her start a conversation with a journal on their AI policies, after receiving vague reviewer feedback generated with AI. Hurdles like these could be significant setbacks without an experienced supervisor, highlighting how in the age of AI, research apprenticeships may be more necessary than ever before.
Supporting a whole ecosystem
Arantes says that AI frameworks need to be supported by best practice policy and signed legislation. She argues there needs to be capability building across the whole higher education ecosystem, not just in research apprenticeships.
There is a future where HDRs are trained to use AI while overcoming risks to critical thinking, judgement and communication. “This requires positive and behavioural changes”, says Zhai.
– Written by Ella McCarthy