From shyness to confidence: Exploring youth agency with an AI-interview practice bot in Kenya

• In Event: AI, Assessment, and Algorithmic Mediation of Learning

Proposal

Abstract
Introduction & Problem Statement
In a world grappling with digital access and unequal exposure, there is a pronounced “AI divide” emerging, with high-income nations disproportionately benefiting from technological advancements while low- and middle-income countries, particularly in Africa, risk being left behind (ILO, 2024). Nowhere is this divide more evident than in the workplace, especially for students in last-mile communities. For them, forming a professional identity and articulating their skills during interviews is a daunting challenge, which often gets compounded by mistrust of the industry and assumptions about fairness in recruitment.
Against this backdrop, 87% of companies are now using AI-driven tools for recruitment (Demand Sage, 2024). For young people in Kenya’s vocational training centers(VTCs), exposure to such tools is not only “novel” but also potentially preparatory for the labor markets they seek to enter. This study re-examines equity and learning by investigating how students interact with an AI-interview bot, a tool that is culturally novel yet accessible via their phones. We conceptualise the bot both as a bridge to equitable opportunity and as a new frontier where issues of trust, fairness, and identity are negotiated.
Central Research Question
This study is guided by a central question: To what extent can a phone-based AI-interview bot reduce the opportunity gap, by offering private, judgment-free feedback on industry expectations, while decentralising access to employability resources, fostering equity across diverse groups of learners.
Methodology
This study adopted a mixed-methods approach to examine how Kenyan youth interact with an AI-enabled interview bot. 50 students from vocational training centres in Nairobi and Kisumu participated, representing diverse trades such as hospitality, automotive, and electrical studies. Each student engaged in a short, in-person facilitated session with the bot, five minutes of interview followed by five minutes of feedback, after which they were encouraged to continue experimenting with the tool independently.
Data collection combined quantitative interaction logs, capturing participation levels and response patterns, with qualitative insights drawn from student reflections, follow-up discussions, and facilitator observations. The bot assesses and provides feedback across five domains: communication, technical know-how, problem-solving, professionalism and attitude, and time management. This design enabled both longitudinal tracking and interpretive depth, meanwhile structured monthly follow-ups over 3 months - until students secure a job, have been designed to support and monitor engagement and evolving perceptions. Together, these methods provided a nuanced picture of the opportunities and limitations of digital interview practice in preparing youth for the labor market.
Preliminary Findings
The pilot revealed five key insights into how youth perceive and work with an AI-enabled interview bot:
Negotiating trust with a novel technology: Initially a bit apprehensive, the students slowly got comfortable as 55% of them progressed from introductory questions into questions on career motivation, training & experience, and aspirations. Only 10% reached the closing probably due to connectivity challenges and the novelty of the experience. The presence of a facilitator eased access, while the bot’s feedback, delivered in encouraging and non-judgmental language, created a safe environment. Unlike their teacher or industry professional who may not be available to provide feedback, the bot offered a proxy environment for rehearsal and growth. Many participants described this as supportive and even companion-like, with some wishing “the bot had a face", signalling their growing acceptance of the tool.
Building confidence through repetition: Several students emphasised that practicing alone with the bot allowed them to gradually shed shyness. As one reflected, “I am not shy now,” suggesting that repeated interaction helped them rehearse and refine self-presentation. The ability to practice privately was cited as a critical feature: “I want access to practice whenever I can and whenever I want.”
Professional identity construction: Students began with modest, hesitant self-descriptions (“I’ve just completed my Grade 3 certificate”) but, over multiple sessions, shifted toward more aspirational narratives (“I want to be the best leader for my team in the kitchen”). The bot’s structured prompts and feedback created a safe space for imagining and articulating a professional future that felt achievable.
Agency asymmetry in interaction: Confident, digitally literate youth “lean in” as they elaborate answers and work on feedback, while less confident youth stay “on script,” giving minimal, ritualistic responses. This asymmetry mirrors broader social divides in digital exposure, confidence, and familiarity with interview norms. So the way forward is to combine in-bot scaffolding with external nudges, via WhatsApp reminders, text messages, in-person visits or follow-up calls, that encourage learners to return, practice regularly, and gradually take more ownership of the interaction.
Substituting for Limited Human Mentorship: Students valued the AI-enabled tool as a meaningful substitute in contexts where teachers or industry professionals were not readily available. They expressed that the bot’s structured questions and feedback gave them confidence to approach real interviews, even without direct human coaching. The ability to receive encouragement, corrections, and practical guidance at their own pace created a sense of being “counselled” despite the absence of mentors. This self-directed support not only expanded their preparedness but also highlighted how technology can fill mentorship gaps in under-resourced training environments.
Discussion & Conclusion
This paper argues that the equity potential of AI lies less in technological sophistication than in its integration into students’ lived realities. For Kenyan VTC youth, the AI-interview bot could serve as both a mirror and a bridge: reflecting fears about interviews and self-presentation while building confidence and readiness for workplaces where such tools are increasingly normalised. By grounding analysis in student voices, we explore how a simple AI-interview bot can help decentralise access to opportunity while cautioning that equity requires making technology a trusted companion in young people’s journeys.

Authors

• Angel Mohan Kandathil, Pratham International

• Sravana Chandra Kalvapalli, Pratham International

• Abhinav Sonone, Pratham International