Why digital classrooms can create new pressures
Schools increasingly use digital platforms to organise teaching, assess progress and support decisions.
Artificial intelligence adds another layer, promising to help teachers interpret information and plan activities.
Yet each new system can also bring training, reporting, troubleshooting and monitoring requirements.
A newly published study suggests that these additional demands deserve as much attention as the potential benefits.
The researchers followed physical education teachers across four measurement waves in China.
Their central question was how institutional digitalisation related to teachers’ stress and subsequent reports of AI-supported practice.
The study found a pattern linking more digitalised institutions with greater reported digital pressure.
That association was weaker for teachers with stronger digital confidence relative to colleagues.
The findings are observational and do not establish that digitalisation itself caused the pressure.
What the researchers studied
Xiangyu Wang and Xiaoyun Wang examined 961 physical education teachers from 120 Chinese institutions.
The researchers combined teacher questionnaires with organisational information about institutional digitalisation.
They distinguished the broader digital environment from each teacher’s own confidence and reported experiences.
The study also measured teachers’ evaluation of AI tools and their later reported involvement of AI in teaching tasks.
The researchers described these measures as separate dimensions of digital work.
A teacher might consider AI useful while still feeling overwhelmed by digital reporting or equipment demands.
Likewise, feeling stressed does not necessarily mean rejecting technology.
The design sought to distinguish between-institution patterns from differences among colleagues within an institution.
That distinction matters because working conditions and individual confidence can operate at different levels.
Four measurement waves
The study collected information at four planned stages during 2026.
The baseline period ran from 7 to 20 January.
Institutional context and digital self-efficacy were measured between 18 February and 3 March.
Teachers’ functional evaluations of AI and their digital stress were measured between 1 and 14 April.
The later measure of AI-collaborative teaching covered 20 May to 2 June.
This ordering allowed the researchers to examine prospective statistical relationships.
It did not make the study an experiment or establish a causal sequence.
The authors explicitly noted that organisational source completion timestamps were unavailable.
Consequently, the planned order of some measurements cannot be treated as definitive evidence of timing.
The main findings
The authors tested five primary hypotheses and applied a correction for multiple comparisons.
Two of the five tests met their corrected significance criterion.
One supported a negative product of coefficients through the digital-stress pathway.
The reported product was minus 0.099, with a pointwise 95 percent confidence interval from minus 0.184 to minus 0.033.
In plain language, the model linked more institutional digitalisation with higher average digital stress, which in turn related to lower later reported AI-supported practice.
This is a statistical association, not proof that one step caused the next.
The second supported result concerned digital confidence and the digitalisation-stress relationship.
The reported interaction coefficient was minus 0.101, with a confidence interval from minus 0.168 to minus 0.037.
The other proposed pathways did not pass the corrected significance threshold.
Confidence helped, but did not remove stress
Digital self-efficacy means confidence in one’s ability to carry out digital tasks.
Teachers who felt more capable than colleagues in the same institution showed a weaker positive association between digitalisation and stress.
That could reflect greater comfort with unfamiliar systems or troubleshooting demands.
However, the positive relationship remained at the low, average and high confidence levels examined.
The results therefore do not support a claim that digital skills completely protect teachers from pressure.
Nor do they show that confidence training alone will resolve organisational workload problems.
The study identifies a relationship that future intervention research could investigate.
Schools may need to consider both teacher capabilities and the demands built into their digital systems.
Why physical education is a useful test case
Physical education teaching is practical, mobile and often time-constrained.
Teachers must supervise moving students, manage equipment and make rapid safety judgements.
Digital tools may assist with movement assessment, feedback and lesson planning.
But devices also require setup, calibration and interpretation.
A teacher who spends additional minutes troubleshooting a system may have less attention available for students.
The study’s AI-practice measure included decision making, prediction, diagnosis, information evaluation and risk recognition.
These are reported activities rather than independently observed teaching behaviours.
The context makes the tension between potential assistance and additional demands especially visible.
The findings should nevertheless not be assumed to apply identically to every school subject.
Digitalisation is not the same as AI use
The researchers used an organisational-source proxy to describe institutional digitalisation.
This measure reflected a broad digital environment rather than the accuracy or quality of one specific AI product.
The later AI-collaborative teaching score came from teachers’ own reports of AI involvement in tasks.
Those two measures answer different questions.
An institution can invest heavily in digital systems without teachers using AI extensively in their lessons.
Conversely, teachers may find useful AI applications even in less digitalised settings.
The researchers tested whether the organisational context predicted later reported AI practice after accounting for baseline practice.
That adjusted direct association did not meet the study’s corrected criterion.
It would therefore be misleading to claim that greater digitalisation automatically increased AI-supported teaching.
Why usefulness and stress can coexist
A common assumption is that useful technology should make work easier.
In practice, a tool can help with one task while creating demands elsewhere.
A teacher might appreciate automated analysis yet struggle with uploading data, interpreting results or meeting new reporting expectations.
The study measured functional appraisal, including perceived adaptability, information quality and personal utility.
It separately measured digital stress, including strain, workload, effort and disruption.
These are not opposite ends of a single scale.
Recognising potential value does not necessarily mean a teacher has enough time or support to use a system comfortably.
The study’s proposed appraisal-related coefficient product did not meet the corrected significance threshold.
That limits conclusions about positive evaluations acting as the main link to later AI practice.
A workplace-design perspective
The researchers drew on the Job Demands-Resources framework.
This framework distinguishes job demands that require sustained effort from resources that help people accomplish work.
Digital systems can potentially provide useful information, feedback and coordination.
They can also increase expectations, interruptions and technical complexity.
The balance may depend on how an organisation introduces and supports its systems.
A technology rollout is therefore more than a software purchasing decision.
It can change who does which tasks and how much time those tasks require.
The new study adds evidence about associations among digital context, reported stress and AI-supported teaching.
It does not test a particular workplace redesign programme.
Why the sample size needs context
The researchers’ source-tracking file contained 1,474 teacher records from 126 institutions.
The final complete-record analysis included 961 teachers from 120 institutions.
That represented 65.2 percent of the source records.
The remaining records had missing-wave, linkage, quality-control or other eligibility and completeness labels.
Some archived exclusion decisions could not be reconstructed from the available files.
The authors also reported that wave-specific response rates could not be calculated.
These limitations matter because teachers who remained in the analytic dataset may differ from those excluded.
The findings most directly describe the included cohort.
A large sample does not by itself remove possible selection bias.
What the study cannot prove
This was an observational study, not a randomised trial.
Institutions were not randomly assigned different levels of digitalisation.
Teachers also reported their own stress and AI-related practice.
Such reports are valuable but may be affected by perceptions, memory and workplace circumstances.
Other unmeasured features of institutions could influence both digitalisation and stress.
The study’s statistical adjustments reduce some alternative explanations but cannot eliminate them all.
A significant coefficient product is not the same as experimentally demonstrated mediation.
The authors were careful to frame their findings as associations.
Claims that AI caused teacher stress or reduced teaching quality would go beyond the evidence.
The implications for schools
The results suggest that schools should examine the workload surrounding new digital systems.
A platform’s promised capabilities do not reveal the full cost of learning and maintaining it.
Administrators could assess whether digital reporting duplicates existing paperwork.
They could also consider whether teachers receive practical technical support.
These are possible management responses, not interventions evaluated in the study.
The evidence does not identify one ideal number of tools or a universal training programme.
What it does support is closer attention to perceived pressure alongside enthusiasm for AI adoption.
Schools may benefit from measuring both technology use and the experience of doing the work.
Future research can test whether reducing unnecessary digital demands improves outcomes.
Why the findings matter outside China
Digitalisation and AI adoption are global issues for schools and employers.
The study’s central distinction between technological opportunity and work pressure is relevant to that wider discussion.
However, its participants were physical education teachers in Chinese institutions.
Local policies, infrastructure and teaching expectations may shape their experiences.
The numerical findings should not be generalised automatically to South African schools or other occupations.
Different environments may have different levels of support and technical reliability.
Research in other countries could examine whether similar patterns emerge.
Comparative studies would also help identify which institutional features matter most.
For now, the results provide a useful hypothesis for wider testing rather than a universal verdict.
How this connects to other AI research
Research Today has previously examined how AI support related to different forms of job redesign among German workers.
It has also covered a study in which ChatGPT peer-review ratings shifted with opinions supplied by users.
Those findings concern different groups and research questions.
They illustrate why AI’s effects cannot be reduced to a single claim about efficiency.
The new teacher study adds the institutional environment and digital stress to that conversation.
It also underscores the difference between using AI and experiencing its implementation as manageable.
Taken together, these articles encourage attention to the full workflow.
They do not establish a common causal mechanism across the separate studies.
What better evidence could look like
Future research could compare schools introducing different kinds of digital support.
Researchers might measure actual time spent on administration, troubleshooting and teaching.
They could also examine whether teacher training changes perceived stress over time.
Independent observation could help validate self-reported AI involvement.
Randomised or carefully designed quasi-experimental studies would provide stronger causal evidence.
Longer follow-up could show whether initial pressure fades as teachers become familiar with systems.
Studies should also assess potential benefits for students, not only teachers’ perceptions.
Understanding both costs and gains would make the evidence more useful for policy.
The current study provides a foundation for these questions.
The bottom line
A four-wave study of 961 physical education teachers at 120 Chinese institutions found a statistically supported stress-related pathway.
More institutional digitalisation was associated with greater digital pressure, which related to lower later reported AI-collaborative teaching.
Teachers with greater digital confidence relative to colleagues showed a weaker digitalisation-stress association.
That relationship nevertheless remained positive across the confidence levels examined.
Three other primary tests did not meet the corrected significance criterion.
The findings are observational and cannot prove that digitalisation caused stress or changed teaching behaviour.
They suggest that schools evaluating AI should look beyond adoption and usefulness.
The everyday demands created by digital systems may be central to whether teachers can use them effectively.
Source Information
Original research: Institutional digitalization and digital stress: a four-wave observational study of teacher-reported AI-collaborative teaching in physical education.
Authors: Xiangyu Wang and Xiaoyun Wang.
Journal: Frontiers in Psychology, published 15 October 2026.
Design: Four-wave observational analysis of 961 physical education teachers at 120 Chinese institutions. Findings are associations, not causal effects.
