How to let students fail – so they can bounce back

In higher education, we often say we prepare students for the real world. Yet, in many classrooms, we continue to design learning environments that minimise or avoid failure, rather than make use of it.

This is a problem.

Outside the classroom, failure is not the exception – it is the norm. Presentations derail, stakeholders push back, questions come from unexpected angles. Success depends not on perfectly rehearsed scripts but on our ability to think, adapt and recover in real time.

• To build resilience, we need to embrace failure in the classroom and beyond
• The power of failure: a case study to foster resilience in students
• How to work smarter, not harder in higher education

If we are serious about preparing students for complexity, we must move beyond performance-focused teaching and start designing for something far less comfortable: productive failure.

How can we create a trampoline-like learning environment, one where students can fall productively, rebound and emerge more confident, capable and self-aware?

In a recent undergraduate module on critical thinking and communicating, I assessed students through a team-based oral presentation. Many of my students were articulate, persuasive and quick to grasp rhetorical principles such as Aristotle’s appeals: logos, pathos and ethos. On the surface, they appeared highly competent communicators.

However, a different picture emerged during mock Q&A sessions.

When I posed probing, unscripted questions to better understand their reasoning, some students became visibly uneasy. They hesitated, overthought their responses and seemed preoccupied with avoiding “wrong” answers. 

The ambiguity inherent in spontaneous questioning, where they could not predict or control the direction of inquiry, created discomfort. It became clear that while they could perform well in structured settings, they were less prepared for dynamic, high-pressure interactions.

This observation led me to rethink how I could better prepare students; not just to perform but to think, adapt and respond in uncertain situations.

Designing for productive failure through AI

To address this, I introduced an AI-enabled platform for formative practice. The platform integrates behavioural science with artificial intelligence to simulate realistic, scenario-based interactions through avatars.

I designed and trained two AI avatars aligned with the types of stakeholders my students would encounter in their graded presentations and, eventually, in professional contexts. These avatars were situated in realistic workplace environments and developed using composite profiles drawn from real-world professionals.

Importantly, the avatars had different personalities and response styles.

It’s not hard presenting and answering questions of those who are open, pleasant and likely to agree. But it’s a different story when the audience does not respond in accordance with what students expected and prepared for.

Students benefitted from practising with both types of stakeholders. Using the Big Five (OCEAN) personality framework, one avatar was characterised by high openness, extraversion and agreeableness, and low neuroticism: representing a receptive and supportive stakeholder. 

The other was intentionally more challenging: high in neuroticism and low in openness, extraversion and agreeableness. This second avatar was more resistant and critical.

Scaffolding practice and building confidence to manage stakeholders

The contrast in avatars allowed for scaffolded practice. Students could first build confidence with a more receptive audience before attempting to engage a more demanding one. In doing so, they developed not only communication skills but stakeholder management strategies, learning to adapt their messaging based on audience profiles.

Low-stakes practice, high-value learning

The impact on student learning was immediate.

Interacting with the avatars was fun! Students described the experience as engaging and even enjoyable, encouraging repeated practice. Crucially, the environment was low-stakes: students could experiment, fail, recalibrate and try again without fear of judgement.

This stood in stark contrast to traditional classroom role-play, which is often limited by time, peer dynamics and variability in feedback quality. Although not a perfect substitute, the AI platform significantly expanded opportunities for deliberate practice. 

Without this formative practice, it would probably mean that much time would need to be set aside for rounds and rounds of role-play with the instructor or peers, with possibly little time for valuable and actionable feedback. While practising with an avatar comes with its inherent limitations, it still provided a platform for increased practice.

Students who had meticulously scripted their presentations found value in verbalising and testing their ideas in real time. The iterative nature of the interactions helped them move beyond memorisation and towards adaptive communication.

Feedback and reflection

Through repeated engagement, students received immediate, contextualised feedback from the avatars. More importantly, they began to reflect on their own performance: identifying what worked, what did not and why.

They quickly realised that a one-size-fits-all approach to communication was ineffective. Instead, they needed to adjust tone, framing and persuasive strategies depending on the stakeholder. This realisation marked a shift from content delivery to audience-centred communication.

Failing forward: a pedagogical imperative

Student feedback revealed something deeper: the AI-enabled practice environment had created a psychologically safe space for failure.

Students were willing to take risks, test ideas and even “fail” in their attempts to persuade the avatars. Yet, through repetition and reflection, these initial failures became stepping stones for improvement. Confidence grew not from getting it right the first time but from learning through iteration.

From a pedagogical perspective, this aligns with the concept of productive failure. By normalising struggle and uncertainty, students learn to navigate complexity, tolerate ambiguity and persist in the face of difficulty. They begin to hold the tension and then reconcile the gap between who they are now and who they aspire to become.

Developing competence, confidence and resilience through failure

Without such tools, providing equivalent practice opportunities would require extensive in-class role-play; often constrained by time and resources. While AI-mediated learning is not without limitations, it offers a scalable way to democratise access to high-quality practice.

With the emergence of educational technologies, educators are better positioned to support students in developing not only competence but confidence and resilience.

Creating environments where students can fail fast and fail forward is no longer optional: it is essential. In a world defined by uncertainty and complexity, the ability to think on one’s feet, adapt to diverse stakeholders and learn from failure could well be the most critical skill we can cultivate.

Yvonne Kong-Ho is assistant professor at the Centre for Professional Communication, Singapore Institute of Technology.

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.