Can AI Promote Self-Efficacy in Learning?

 

Can AI Promote Self-Efficacy in Learning? 

Artificial Intelligence (AI) is fast becoming a classroom companion, helping students write essays, solve equations, and even rehearse speeches. Nevertheless, amid the hype and hesitation, an important question arises: Can AI help learners believe in their own abilities?

Central to this question is the psychological concept of self-efficacy — the learner’s belief in their ability to succeed in a specific task. According to Albert Bandura (1997), self-efficacy drives motivation, persistence, and achievement. When students believe they can learn, they are more likely to do the hard work of learning. In an AI-driven world, the challenge for educators is ensuring that AI, a tool capable of both empowering and disempowering, strengthens that belief rather than replacing it. Building self-efficacy is not just a goal; it is a necessity for future-ready learners.

What Is Self-Efficacy — and Why It Matters

Self-efficacy is not the same as general confidence; rather, it refers to the specific belief in one's abilities, such as the conviction of "I can learn this." Research by Bandura (1997) indicates that students with high self-efficacy are more likely to engage in deeper learning, recover more quickly from setbacks, and exhibit greater resilience when facing challenges.

For decades, teachers have built self-efficacy through four main pathways:

1. Mastery experiences — succeeding through effort.
2. Vicarious experiences — observing peers or models succeed.
3. Verbal persuasion — encouragement and constructive feedback.
4. Emotional regulation — reducing anxiety and cultivating calm focus.

The exciting question is how AI tools can contribute meaningfully to each of these. In other words, can the machine help the human regain their belief in their own capacity to learn?

1. Mastery Experiences Through Personalised Feedback

The greatest potential of AI to enhance self-efficacy lies in its ability to provide timely, adaptive feedback. In traditional classrooms, feedback often comes with a delay; assignments may be returned days later, leaving students uncertain about their progress. AI systems can fill this gap by offering immediate responses that recognise achievements and guide students on their next steps.

Adaptive learning platforms such as CenturyTech, Khanmigo, and Squirrel AI offer immediate feedback that recognises progress and suggests next steps for learners. When a learner receives evidence of their growth, such as “Your sentence variety has improved by 15% this week!” they experience a significant moment of mastery. These small wins become evident, reinforcing the belief that effort leads to improvement.

Research supports this effect. Studies in AI-supported tutoring environments have shown measurable gains in self-efficacy when learners receive personalised, mastery-oriented feedback (Keller et al., 2021; Roll & Wylie, 2016). The key, however, is tone and framing. AI feedback that emphasises growth rather than deficiency cultivates persistence rather than perfectionism.

“Students need to feel that AI is noticing their progress — not grading their worth,” says Dr Maria Keller, an educational psychologist who studies AI motivation models.

2. Verbal Persuasion and Encouragement

AI chatbots, virtual tutors, and tutoring assistants can effectively enhance verbal persuasion by providing feedback that encourages effort and perseverance. Simple responses like, “You’re close! Try adjusting your phrasing,” or “That’s a creative approach — let’s explore it further,” can have a surprisingly strong psychological impact.

Although AI lacks genuine empathy, well-designed systems can simulate motivational dialogue that mirrors the language teachers use. This can be especially beneficial for students who are shy or self-conscious, or who study independently.

However, this is also where the human element remains essential. Students must learn to interpret AI’s “encouragement” critically — to see it as guidance, not authority. Teachers can model reflective prompts such as:

• “Do you agree with that feedback?”
• “How might you adapt or challenge what AI suggested?”

This interplay helps learners internalise self-persuasion — the real goal of self-efficacy development.

3. Vicarious Learning in Virtual Environments

Another way to build self-efficacy is through observing others succeed, known as vicarious experience. AI-driven simulations and digital mentors can provide virtual environments where students can witness effective problem-solving and creative performances.

Although AI lacks genuine empathy, well-designed systems can simulate motivational dialogue that mirrors the language teachers use. This can be especially beneficial for students who are shy or self-conscious, or who study independently.

For example, an AI drama coach might model how to build dialogue tension, or a science simulation might demonstrate how to troubleshoot an experiment. These “digital role models” can demystify complex skills, showing that success is attainable through strategy and reflection, not just innate talent.

A study by Luo, Liu, and Wang (2023) found that students who practised public speaking with AI-driven rehearsal software not only reduced anxiety but also reported stronger self-efficacy in communication tasks. The key was repeated exposure to success scenarios — practising until they could see and feel improvement.

Still, balance matters. Teachers should ensure that AI models reflect the diversity of voices, bodies, and perspectives — so that all learners can see themselves represented in examples of success.

4. Emotional Regulation and Safe Practice Spaces

One of the most overlooked advantages of AI in education is its ability to create low-pressure practice environments. Learners can rehearse, experiment, and make mistakes privately, without facing social judgment. For students who experience anxiety, have language challenges, or are neurodivergent, this sense of psychological safety can be genuinely transformative.

AI writing assistants, for instance, can provide supportive critique without the sting of public correction. Similarly, AI performance coaches in music or theatre can offer immediate technical guidance without embarrassment. As one student remarked after using an AI acting coach:

“It’s like practising with a mirror that talks back — it tells me what to fix, but it never laughs.”

Reducing emotional barriers, such as fear and shame, directly supports Bandura’s fourth pathway to self-efficacy: regulating affective states to enhance focus and persistence.

5. Supporting Metacognition and Reflection

AI can serve as a metacognitive mirror, helping students reflect on their thought processes. By posing questions like, “Why did you choose this argument?” or “How might your audience respond to this?” AI encourages learners to articulate their strategies, assess outcomes, and plan for improvements.

This reflective process enhances metacognitive awareness, a crucial factor in academic self-efficacy (Zimmerman, 2000). When students can explain their learning methods, they start to perceive themselves as active participants in their own educational journey.

However, reflection must remain authentic. If AI provides every insight, students risk parroting rather than processing. Educators can safeguard metacognition by designing AI-assisted reflection journals in which learners critique the AI’s suggestions—evaluating which insights feel genuine and why.

The Double-Edged Sword: When AI Undermines Self-Efficacy

Although AI has significant potential, it can also undermine self-efficacy when misused. The qualities that make AI beneficial—such as automation, accuracy, and speed—can leave learners feeling obsolete. Educators play a vital role in guiding the use of AI to prevent dependency and bias, ensuring that AI remains a tool for empowerment instead of a crutch.

Over-Scaffolding

When AI performs too much cognitive work, students lose ownership of learning. If an AI writes most of the essay or solves every equation, learners see themselves as passive recipients rather than capable doers.

Over-Confidence and Dependence

Conversely, students who lean on AI for every task may develop false self-efficacy — confidence unanchored to genuine skill. This “illusion of mastery” crumbles when the AI is unavailable, leading to frustration and disengagement.

Algorithmic Bias

AI systems that are trained on biased data can deliver unequal feedback. If specific writing styles or dialects are unfairly labelled as “poor quality,” students from marginalised linguistic or cultural backgrounds might start to internalise negative beliefs about their abilities.

To address this issue, we need pedagogical mediation. This involves teachers guiding students in interpreting, critiquing, and contextualising AI feedback within a human framework of understanding.

Educators as Self-Efficacy Architects

Teachers are the ultimate architects of self-efficacy. AI serves as a new set of tools in a teacher's toolkit, not as a replacement for their skills. The teacher’s role evolves to design learning experiences that support student agency rather than undermine it.

Here are five strategies educators can use to ensure AI strengthens self-belief rather than diminishes it:

1. Set Transparent Goals: Before introducing AI, help students define what success looks like — “By the end of this task, I want to be better at...”
2. Encourage Reflection: Use AI chatbots for reflective questioning, but have students journal their responses in their own words.
3. Celebrate Growth: Highlight how students improved with AI’s help — but emphasise that they made progress.
4. Demystify the Algorithm: Explain how AI generates feedback. When students understand the process, they retain agency.
5. Promote Ethical AI Use: Discuss authorship, originality, and when AI assistance becomes dependent.

Beyond the Classroom: Building Future-Ready Learners

In an AI-rich world, self-efficacy is not a luxury — it’s a necessity. As technology automates routine tasks, human learners must bring curiosity, adaptability, and reflective confidence.

AI’s real educational promise lies not in producing perfect answers, but in helping students build the courage to ask better questions. The ultimate measure of success is not how fluently students can use ChatGPT or Claude, but how confidently they can say, “I can learn this — with or without AI.”

References (APA 7th)

Aleven, V., Roll, I., McLaren, B. M., & Koedinger, K. R. (2017). Help helps, but only so much: Research on metacognition and tutoring. Educational Psychologist, 52(4), 284–311.

Bandura, A. (1997). Self-efficacy: The exercise of control. W.H. Freeman.

Keller, J. M., Suzuki, K., & Krosnick, J. A. (2021). Adaptive systems and motivation: Applying the ARCS model in digital learning environments. Springer.

Luo, T., Liu, Y., & Wang, J. (2023). Virtual rehearsal and AI coaching in public speaking: Reducing anxiety and improving self-efficacy. Computers & Education, 199, 104755.

Roll, I., & Wylie, R. (2016). Evolution and revolution in artificial intelligence in education. International Journal of Artificial Intelligence in Education, 26(2), 582–599.

Zimmerman, B. J. (2000). Self-efficacy: An essential motive to learn. Contemporary Educational Psychology, 25(1), 82–91.