Doctoral Education Meets the AI Revolution: How PhD Candidates Are Affected

To shed light on the issue, SLU and the Sparbanken Skåne Centre for Sustainable Primary Production invited PhD candidates and supervisors to an afternoon seminar on the role of AI in doctoral education. The centre works to promote knowledge-based, sustainable and resilient primary production, and the eight PhD candidates linked to the initiative cover the full breadth of the food system – from genetics and plant breeding to urban farming, governance and energy integration. Together, they contribute new perspectives on how future sustainable food production might take shape.

On the way to completing a thesis, PhD candidates inevitably face the question of how AI should be used. To make the most of these tools, it is essential to understand how generative AI works and what limitations it has.

– To get the most out of it, you need to know how to formulate a prompt – although you can ask AI for help with that as well, says Matt Low, AI trainer at SLU and the event’s keynote speaker.

Generative AI is more reliable when working in areas with extensive existing data. At the same time, it is crucial that candidates critically assess the responses they receive, as AI often expresses itself with great confidence and sometimes with inherent bias.

– If you ask a vague question, you will get a vague answer, says Matt Low. Be detailed, be specific, and above all: keep asking questions and do not accept all answers as true.

AI tools can support idea development, act as a sounding board, or assist with practical tasks such as text editing. However, its use must be anchored in an open dialogue between candidate and supervisor. Both parties need to agree on how AI may be used to ensure that the candidate’s authorship is not undermined.

Doctoral studies also come with the expectation that certain learning processes must be completed. Supervisors need a clear view of which tasks they expect the candidate to carry out independently, and should not be left at the end of the process feeling that the thesis has been produced through shortcuts.

– Students have always received help in various ways, says Matt Low. The question is how much, and what kind of external assistance, is acceptable?

In addition to pedagogical considerations, there are ethical aspects. Sharing personal data or other sensitive information with AI tools can be problematic. There are ways to work with generative AI safely, but doing so requires deliberate choices.

– AI should function as a tool that enhances – but does not replace – critical thinking, analysis and argumentation, says Matt Low. The candidate must be able to explain and defend their perspectives, data, analysis and conclusions.