Meet the first AI-native university cohort

The call to invest in formal training

Universities must support faculty through this change, Professor Ngonzi states, as it is not easily solved individually, especially when the questions involve privacy, accessibility, academic integrity, intellectual property and equity.

“Faculty need time to redesign assignments and somewhere to turn when they are unsure what is appropriate in their discipline,” she adds.

The rise of AI is reminiscent of the spring of 2020 when the pandemic hit. As the world switched to digital, many universities asked faculty to move their entire curricula online with just a few days of notice – something which understandably caused huge backlash.

Institutions had to find ways to support their staff: providing software and online teaching training, identifying peer mentors among the faculty who could guide others and developing new policies around remote assessments.

“We are in an almost identical moment today with the arrival of AI-native students, and it is going to require the exact same level of institutional support,” she explains, adding that it isn’t enough for universities to just issue vague academic integrity statements. They must instead invest in formal, role-specific retooling.

That support will need to extend beyond simply helping faculty police AI use. One of the ways AI can help is redesigning the curriculum itself, offloading the mechanical, repetitive parts of teaching and freeing up professors to focus on mentorship.

It can be used to meet students where they already are – customising explanations, formats and pace to match an individual’s learning style and interests.

“We stop forcing everyone to fit into a single, standardised mould and instead use the technology to normalise excellence,” says Professor Ngonzi. “That is the real support faculty need — the permission, the compensation and the training to let go of legacy grading models so they can focus on the human interactions where real learning actually happens.”

Instead of reacting to advancements in AI, universities must instead move to proactive pedagogical redesign, agrees Professor Alison Gibb, Director of Learning, Teaching and Scholarship at Adam Smith Business School in the UK.

“We must invest in staff development and redesign assessments so that AI use is either meaningfully integrated or deliberately excluded where unaided human capability is the learning outcome,” says Professor Gibb.

It’s likely that this generation of learners will be comfortable moving between human and AI work, but some will struggle to deviate from what they already know the tool can do, particularly when they’re under time pressure or carrying out unfamiliar tasks.

“Their technical fluency will be high, but their critical and ethical literacy around citation, bias, accuracy and responsible use will be uneven and untrained, so we will need to ensure that programmes explicitly teach and assess these skills,” she adds.

Professor Glib also raises the point that access to premium tools and reliable devices may create an “AI divide” that could disadvantage students who only have access to the free ‘lite’ versions or have not had the opportunity to experiment with prompting and evaluation.

To target that, Professor Gibb says business schools and universities must establish a school-wide approach to AI in learning and assessment, so they avoid contradictory expectations between courses about what AI use is allowed, expected or prohibited.

“Programme directors should articulate a short, shared statement on where AI is expected and where it is permitted or prohibited,” she says.

Challenge students in ways that truly test their abilities

Instead of putting out blanket restrictions on AI, course design needs to integrate more oral or in-class presentations of ideas to verify that students understand the material, explains Professor Gibb. Reflective components should be built in so that students can explain their process and how and why they used AI.

“We don’t expect a uniformly weaker cohort, but rather one where weaker students can, at first glance, produce good work while excellent students will need to work harder to stand out. It’s up to us to keep challenging them in ways that truly test their abilities and prepare them for an unpredictable tomorrow,” says Anna Plechatá Krausová, Chancellor of NEWTON University in Czechia.

One benefit of the Czech education system is that exams are largely based on oral examinations, including an in-person thesis defence across all educational levels. In-person assessments enable academics to assess the students’ process, not just the finished work.

Other ways to test learning could be through live problem-solving, live case analysis, simulations, writing retreats and practical projects.

But testing students without AI doesn’t mean that it should be removed completely from the curriculum. Instead, universities could use the technology to make learning more challenging. They could, for example, use AI to complicate the task, come up with new simulations or play the devil’s advocate.

“In Plato’s cave, people mistake shadows on a wall for reality because they have never seen anything else. AI can produce those shadows in perfect resolution, personalised precisely to our preferences. Because the answers are persuasive and comfortable, we may never think to turn towards the light,” says Dr Krausová.

In other words, if academics are not given the guidance, time and training to evaluate AI-supported academic output, they risk losing sight of the main goal, which is to make students turn towards evidence, reality and independent judgment.

Lean into ‘building with machines’

For some academics, however, the answer is not to make students work around AI, but to make the technology part of increasingly ambitious work.

Aron Lindberg, Associate Professor of Information Systems at US-based Stevens Institute of Technology, believes we should “lean into building with machines”.

As students will inevitably use AI, academics must ask for increasingly complex outcomes.

“Everything else should get roughly ten times more ambitious,” he says. “If students have AI agents, the project should involve real software, a real client or a local organisation, and messy human requirements and expectations.”

In practice, this means putting capstone projects first in the curriculum, rather than last. These capstone projects, where students collaborate on a real business challenge, allow students to begin their learning with hands-on work, starting with an actual problem in week one and building with AI for the whole programme, keeping humans at the centre of the process.

“I think we should be experimenting with the model of higher education itself rather than defending it,” he says. “Knowledge transfer has become cheap, and what's left is character formation, helping students become the kind of people who reach for hard problems and start building solutions.”

Another way that the Stevens Institute of Technology is leaning into this is by building a new undergraduate major in Business and AI, complete with a “build studio” – a place for students to engage with stakeholders to build solutions that address important problems. The goal is to teach students not how to prompt, but how to manage agents and implement AI in an organisation.

“They don't produce a first draft on their own and then check it against a model; they start with the model. I don't think universities should fight that,” says Professor Lindberg. “What knowledge is needed is determined by what you are building, rather than the other way around, so what students need to learn gets pulled in as the project demands it”.

Don’t use AI for every aspect of learning

However, there is a danger in assuming that because students are comfortable using AI, they should use it for everything.

When this year’s cohort enters the classroom, they will have already used AI, but in a fairly limited capacity, says Rodrigo Belo, Professor of Operations, Technology and Innovation Management at Nova School of Business and Economics in Portugal. His concern for this cohort is that they will automatically turn to AI to answer a question rather than trying to figure out the answer first.

“We need to encourage students not to use AI for everything, and to spend time trying to reach answers themselves,” he says.

It’s important to teach students when to use AI as well as how to use it, he adds. For example, AI can be used more productively when you understand the problem you are trying to solve and can critically evaluate the answers it gives you.

Yet, there are certain things that Professor Belo believes AI should not be used for.

“I allow students to use AI for almost everything except writing: they should write their own sentences. Writing creates friction that is useful. Feeling stuck is often a sign that there is something we still don’t understand,” he says. “If AI removes that effort, it can also remove the learning.”

One way to preserve this is by making courses “AI-proof” – designing learning activities and assessments in ways that preserve the effort, thinking and interaction that learning requires. It is there that interaction in the classroom becomes even more important.

However, he warns schools not to move too quickly.

“AI is an amazing technology with great potential, but some schools may be going too fast and may have to backtrack. Universities should experiment with AI, but should be careful not to go too fast,” he says.

Students are not AI natives… but neither are faculty

All of this points to an important caveat: familiarity with AI does not necessarily make students expert users of it.

As Professor Belo mentions, while this new cohort will have experience using AI on a simple level, this does not mean that they will be competent users of the technology.

“Personally, I think it is still in its infancy with regard to the upcoming cohort,” says Dr Michael Drummond, Programme Leader at the UK’s Liverpool Business School. “I suspect there has been little to no guidance on appropriate use of GenAI. This therefore presents an opportunity for institutions to refine this in preparation for graduation.”

“Since AI is not yet part of the Key Stage Curriculum, we are not yet at a stage where we can call this generation GenAI native,” he explains. Yet Dr Drummond believes they are still digital natives, so their relationship to new technology is less risk-averse than previous generations.

Equally, many academic faculty members are not necessarily fluent in AI yet. “My experience is that some staff are still finding their way around the technology,” says Dr Drummond. “They may not feel able to provide concurrency to their subject area as the technology is evolving quickly. Many staff want to understand the ‘ins and outs’ before adding it to their curriculum.”

Despite this, Dr Drummond doesn’t believe the approach to teaching should change.

“We should always be current in our subject area, and GenAI is just another addition. So, rather than different, I would see us adapt to incorporate GenAI use both in teaching and assessment in how it can be a supportive tool and not a replacement,” he says.

He emphasises that traditional assessments need to be revised with GenAI in mind, to incorporate how to use it to study, and also understand how it is used in industry. This will help prepare this next cohort for the world they’ll be entering after graduation.

Perhaps that is a more useful way to think about the cohort entering university now. They may be among the first students with years of experience using AI, but experience is not expertise. Universities will need to teach them not only how to use these tools, but when to question them, when to put them aside and what kinds of thinking should remain their own.

What’s certain is that the world these students will graduate into will be very different to the one we see today. With AI advancing rapidly, students – and faculty – must try to keep up with the pace.