Does AI dependence mediate the link from academic stress to burnout and anxiety in the same way for both outcomes?

Why one habit could feed both burnout and anxiety

The proposed explanation is about competence, not screen time. When academic stress rises, students may lean on generative AI as a coping tool; the study describes this as maladaptive dependence rather than ordinary use [1]. The concern is that outsourcing thinking reduces the small successes that build a sense of mastery, so students come to rely on outside support and trust their own ability less [1].

That weakened self-efficacy is the hinge. A student who feels less capable of handling the work has two natural reactions: exhaustion and detachment from study, which is the burnout side, and worry about performance, which is the anxiety side [1]. Because both outcomes sit downstream of the same eroded confidence, the same mediator can plausibly feed both.

Same shape does not mean identical strength

The study surveyed 1,623 college students and used a statistical mediation model that traced stress through AI dependence and self-efficacy to each outcome separately [1]. That design can show that a pathway exists for burnout and that a pathway exists for anxiety, but testing them separately does not establish that the two pathways are equally strong or that one outcome is more affected than the other [1].

The authors frame the result as a sequential mediation: stress, then AI dependence, then reduced mastery and self-efficacy, then burnout or anxiety [1]. The word "sequential" matters here. It means the effect is described as running through a chain of steps rather than AI dependence acting alone, so any claim that dependence by itself explains the link would overstate what was tested [1].

What the surrounding studies do and do not add

A separate line of work looks at a different AI-related feeling: anxiety about being replaced by AI. A 10-day diary study of medicine and nursing students tracked how AI replacement anxiety, state anxiety, and academic burnout indicators moved together over time [2]. That study did not include academic stress as a starting point, so it cannot confirm or contradict the stress-to-burnout pathway described above [2].

Another study examined how academic stress relates to dependence on AI-generated design tools among design students, with procrastination and help-seeking as intermediate steps [3]. It did not test burnout or anxiety as outcomes, so it speaks to how AI dependence may form but not to whether it mediates the same way for both outcomes [3].

Taken together, the direct evidence for the original question comes from the mediation study alone [1]. The other two studies are consistent with the idea that AI-related reliance and AI-related worry are both tied to student distress, but neither tests the specific comparison between burnout and anxiety [2][3].

What this means for how students use AI

The practical implication the authors draw is not to avoid AI but to use it differently: as a scaffold that supports a student's own thinking rather than a substitute for it, with reflective and autonomy-supportive use encouraged [1]. That recommendation follows from the self-efficacy explanation, since the proposed harm runs through lost mastery experiences rather than through AI use in general [1].

One caution belongs beside the result. The mediation study is cross-sectional survey data, so the ordering of stress, dependence, self-efficacy, and outcomes is modeled by the researchers rather than observed over time [1]. The diary study does track day-to-day associations, but for AI replacement anxiety rather than academic stress [2]. Whether heavier AI dependence actually precedes and produces later burnout and anxiety, and whether it does so equally for both, remains to be tested with longitudinal designs.

Sources used in this answer

Related research questions