AI, pedagogy and the risk of ‘smoothing’: Vanessa Camilleri on what may be lost

As artificial intelligence becomes more embedded in education, the conversation often centres on efficiency, assessment, and integration. But for Vanessa Camilleri, an Associate Professor at the Dept of Artificial Intelligence, University of Malta, the more pressing question is what these systems are quietly reshaping in the process.

“There is an underlying expectation that learning can be designed, structured, and evaluated,” she says. “Yet within this framing, there is also an implicit narrowing of what is recognised as learning.”

Speaking after attending Vanishing Acts: AI, Performative Knowledge & Sustainable Memory, Prof. Camilleri reflects on a recurring idea that stayed with her: a “smoothing” effect brought about by AI. As processes become more optimised, she notes, something less visible begins to shift. “The increasing smoothness of production processes may come at the cost of texture, difference, and the distinctiveness of creative expression.”

This is not a rejection of AI, but a reframing of its impact. Rather than asking whether the technology works, discussions at the conference focused on how it alters the conditions under which decisions are made. “The discussion did not centre on whether these systems function effectively,” she explains, “but on how they influence what is perceived as reliable, and how they participate in the construction of reality itself.”

Rethinking how learning happens

For Prof. Camilleri, this has direct implications for pedagogy. If learning continues to be defined by what can be measured and optimised, other forms of knowledge risk becoming harder to recognise.

“In many of the sessions, learning appeared not as a linear progression from objective to achievement,” she says, “but as something that unfolded through interaction, uncertainty, and ambiguity.”

Conversations with practitioners, including artist Vince Briffa, reinforced the idea that AI should be approached as a material rather than a solution. “The question was not whether AI should be embraced or resisted, but how it might be engaged with,” she notes.

What emerges is a shift away from seeing AI as simply a tool to improve existing systems. Instead, it points to a need to reconsider the systems themselves. “We are not simply being asked to integrate new tools into existing educational models,” Prof. Camilleri says, “but to rethink the conditions under which learning takes place.”

Her reflection ends without a definitive conclusion, but with a clear warning. “As we make learning more efficient and more aligned with systems we can manage, we may also be narrowing the space within which learning can occur,” she says.

“Not everything that matters in learning can be anticipated, and not everything that cannot be anticipated should be allowed to fade from view.”