Learning by doing in a GenAI-enabled world

For decades, we’ve paid lip service to the idea that students learn best by doing, even as our magisterial lectures saw engagement dwindle and attention fragmented by digital distractions (like shopping online for a shirt). 

I can’t do lectures. I’ve never been able to. I get very distracted by disengaged students – I can hardly keep myself from stopping the whole class to ask about the colour of the shirt. So, my own response has been to build project-based subjects that put the work of learning directly in the students’ hands. This makes it harder to quantify learning by a final exam, but it transforms education from memorisation to the cultivation of enduring and life-applicable skills.

However, today we have to bring GenAI into our classrooms – our graduates must be GenAI-savvy. More crucially, we have to ensure the students remain the primary source of creativity and critical thought in the process – knowing how to obtain the best answers from GenAI tools, but also how to scrutinise the outputs. What does this mean for learning by doing?

• Why GenAI helps some students but not others (and what to do about it)
• ‘Your bot of choice is not a filing cabinet’
• Conversations with bots: teaching students how – and when – to use GenAI for academic writing

One of my subjects is a sociolinguistics module, Language Matters. It is not only project-based but also a Collaborative Online International Learning initiative, run in partnership with the University of Seville in Spain, under the Explore, Create and Offer (ECO) project. Students in Hong Kong and Spain jointly select a sociolinguistic topic, identify a real community issue (the Explore phase), and work towards a tangible solution (Create) to be presented back to that community (Offer). 

This model – applied to disciplines from ergonomics to linguistics – thrusts students into authentic engagement, requiring them to interview industry experts, conduct surveys with stakeholders and grapple with the difficulties faced by the people affected, often highlighting their own misconceptions.

From day one, this pushes students far beyond their comfort zones. A pervasive initial doubt is: Can I actually do this? For most, this immersive, self-directed approach is entirely new. My first objective is simply to show them that they can. We start, just in Hong Kong, with a compressed, two-week mini-pilot project where students explore a self-chosen topic and produce an academic poster – another first for many – with encouragement to present the findings at undergraduate conferences.

To spark immediate engagement, I begin with humanity. I use powerful photographs by an artist formerly based in Hong Kong, Raúl Hernández, who focuses on marginalised communities: Filipino and Indonesian domestic workers and the Pakistani minority. By asking students to study these images – to read the stories in the context of a scene – sociolinguistics stops being abstract. Issues of language, power and identity become real, providing fertile ground for their own project ideas.

The second, often surprising, pedagogical tool I introduce is GenAI itself. For many, this is the first time an educator has actively encouraged its use. Their initial hesitancy is palpable: Is this allowed? How do I even start? GenAI has a parallel data-processing capacity no human can match. Given it is here to stay, our duty is not to ban it, but to train students to use it ethically and effectively. 

In Language Matters, students must identify theoretical frameworks for their research. For those without a sociolinguistics background, GenAI is an entry point that reveals theories and paradigms they would not have encountered otherwise. As their topics stray into other disciplines, like marketing, media studies or health-communications, I often end up learning these paradigms with them. This means that most students will not read every paper in full – but let’s be honest, how many do, even without GenAI? 

The true test comes in the project’s creative, solution-building phase. Here, GenAI recedes to a secondary role. The core work must be driven by what students discover in their community, through conversations with experts and the people they are trying to help. GenAI cannot judge feasibility or cultural fit. One student group, for instance, identified a lack of medical information for Indonesian domestic workers in Hong Kong and developed an informational website, only to discover that their target community did not seek information this way. The solution – a Facebook page, the platform the community actually used – emerged not from an algorithm, but from empathetic listening and adaptation.

The ultimate challenge, entirely beyond a machine’s reach, is the human collaboration itself. My students must navigate a seven-hour time difference, academic and professional workloads (because many students in Hong Kong and Seville have jobs as well), and subtle cultural nuances with their Spanish peers. This complex dance of negotiation, patience and mutual understanding is probably the hardest part of the subject and something they must experience in order to learn. Through practice, missteps and resolution, they build the intercultural competency that is the hallmark of a global graduate. They learned it by doing it.

They tell me this subject is the hardest they have ever taken. They also tell me it is the best. In a world being transformed by GenAI, it is precisely this arduous, people-centric learning – the kind that forges critical thinkers, creative problem-solvers and empathetic bridge-builders – that will remain indispensable. Our task is to design the classrooms where these irreplaceable skills are developed. 

Renia Lopez-Ozieblo is assistant dean at the Faculty of Humanities at The Hong Kong Polytechnic University. AI was used to polish the language of the article.

If you would like advice and insight from academics and university staff delivered direct to your inbox each week, sign up for the Campus newsletter.