Artificial intelligence is becoming part of ordinary work, but organisations face a harder question than whether employees have access to the technology: do they understand it well enough to use it productively?
A one-year longitudinal study published on 1 October 2026 suggests that AI literacy may matter for how engaged employees remain at work. Researchers Qingqi Liu, Yuju Lei and Jingjing Li followed 1,108 employees aged 21 to 60 across three survey waves and found that higher AI literacy at the start of the study predicted stronger work engagement one year later, even after accounting for employees’ starting level of engagement and several demographic and technology-use factors.
The result is particularly relevant as employers move beyond experimental AI projects and embed generative AI, automated decision support and other intelligent tools into everyday workflows. The findings indicate that simply putting AI in employees’ hands may not be enough. The knowledge and confidence needed to understand and work with these systems could shape whether AI becomes a useful workplace resource or another source of strain.
Following employees for a full year
The study used a three-wave longitudinal design rather than a single snapshot. At the first measurement point, the researchers recorded demographic information, AI literacy and work engagement. Six months later, participants were assessed on work self-efficacy and job burnout. At the final wave, one year after baseline, work engagement was measured again.
This sequencing allowed the researchers to test whether earlier AI literacy predicted later engagement and whether intermediate psychological factors helped explain the relationship. The analysis controlled for gender, age, intensity of AI use and baseline work engagement, reducing the likelihood that the final association simply reflected employees who were already more engaged or who happened to use AI more frequently.
AI literacy at baseline significantly and positively predicted work engagement at the one-year follow-up. Importantly, the association remained when the researchers introduced the proposed mediating variables into the model.
Confidence and burnout helped explain the link
The analysis identified three indirect pathways between AI literacy and later work engagement. Employees with stronger AI literacy tended to show higher work self-efficacy six months later, and this was associated with stronger subsequent engagement. AI literacy was also linked to engagement through lower job burnout.
A third pathway connected the two mechanisms in sequence: AI literacy predicted work self-efficacy, which was associated with burnout, which in turn was related to later work engagement. The pattern suggests that the value of AI literacy may extend beyond knowing which button to press or how to construct a prompt. Employees who understand AI may feel more capable of handling technology-rich work, while that stronger sense of capability may help protect against exhaustion and disengagement.
That distinction matters because AI adoption can create competing experiences inside the same organisation. A capable user may see an AI system as a way to remove repetitive work, explore alternatives or complete difficult tasks more efficiently. Someone who does not understand the system’s limitations may instead experience uncertainty about when to trust it, how to verify its output or whether their own skills are becoming less relevant.
The study does not establish that improving AI literacy will automatically cause engagement to rise. Its longitudinal structure strengthens the temporal evidence, but the measures are still observational. Employees were not randomly assigned to AI-literacy training, so unmeasured differences between workers could still contribute to the relationships.
Why this matters for employers
Many workplace AI programmes are still measured through adoption statistics such as licences activated, prompts submitted or hours saved. Those indicators reveal whether a tool is being used, but they say relatively little about whether employees are becoming competent and sustainable users of it.
The new findings point toward a broader implementation question. Training that focuses only on tool mechanics may miss the capabilities employees need to judge AI output, understand appropriate use cases, recognise errors and maintain confidence in their own contribution. If AI literacy supports self-efficacy while reducing burnout, those human outcomes may be as important to long-term adoption as technical usage itself.
For South African employers, the issue is especially practical as AI capability develops unevenly across occupations and organisations. Two employees may have access to the same generative AI system while possessing very different levels of confidence, critical evaluation ability and understanding of the technology. Treating access as equivalent to literacy risks obscuring that gap.
The findings also caution against interpreting more AI use as necessarily better. The researchers controlled for AI-use intensity, meaning the association between literacy and later engagement was not simply a reflection of frequent users reporting better outcomes. Knowing how to work with AI and merely spending more time using it are not the same thing.
Important limitations remain
The sample consisted of employees surveyed in a specific research context, so the strength of the relationships may differ across countries, industries and organisational cultures. Work engagement, self-efficacy and burnout were measured through survey instruments, which can be influenced by common response tendencies. The design also cannot eliminate all alternative explanations or prove causality.
Further research could test whether structured AI-literacy interventions produce measurable improvements in self-efficacy, burnout and engagement compared with control groups. It would also be useful to distinguish between different forms of AI literacy, including technical knowledge, critical evaluation, ethical understanding and practical task competence.
Even with those caveats, the one-year design offers a useful counterpoint to the idea that workplace AI success is mainly about acquiring the newest tools. The evidence suggests that what employees know about AI, and how capable they feel while working with it, may influence whether the technology becomes part of an engaged working life rather than simply another system employees are expected to use.
Source Information
Study Title: Does AI literacy enhance employee work engagement by boosting work self-efficacy or reducing job burnout? A one-year longitudinal study
Authors: Qingqi Liu, Yuju Lei and Jingjing Li
Journal: Humanities and Social Sciences Communications
Year: 2026
DOI: 10.1057/s41599-026-09184-7
