A new scale separates feeling that learning is redundant from feeling self-driven, and it works similarly across student groups.

Researchers validated a new survey tool to measure how generative artificial intelligence relates to students’ motivation and basic psychological needs in higher education. The scale showed a clear multi-factor structure and worked comparably across gender, study level, field, and frequency of artificial intelligence use. Higher artificial intelligence tool use was linked to stronger beliefs that learning tasks are redundant and to slightly more controlled motivation.

Quick summary

• What the study found: The AI-Motivation and Needs scale reliably captured two types of redundancy beliefs and three types of motivational orientation, and heavier artificial intelligence use correlated with more redundancy beliefs and more controlled motivation.
• Why it matters: Schools can measure whether artificial intelligence is supporting learning or quietly shifting students toward “just get it done” engagement.
• What to be careful about: Some subscales had modest reliability and several key concepts were measured with single items, limiting precision.

What was found

In the journal article Validating the AIM–N: An AI-motivation and needs scale with multi-group invariance and MIMIC-DIF evidence in higher education, researchers introduced the AI-Motivation and Needs scale to assess how artificial intelligence integration relates to students’ motivation and need satisfaction.

Survey data from 904 university students supported a five-factor structure: two artificial intelligence-related redundancy belief subscales (task-level and motivational or identity-level) and three artificial intelligence-related motivational orientations (intrinsic, identified, controlled). Model fit was acceptable, and redundancy items showed strong internal consistency.

On average, students reported low-to-moderate redundancy beliefs. Task-level redundancy averaged 1.92 out of 5, while motivational or identity redundancy averaged 2.19 out of 5.

Identified motivation had the highest mean (5.88 out of 7), intrinsic motivation was moderate (4.77 out of 7), and controlled motivation was lower (3.13 out of 7). The single amotivation item (“Why do it myself if AI can do it?”) showed wide spread: about a quarter agreed while half disagreed.

What it means

The key psychological signal is not simply “students use artificial intelligence,” but how they interpret what that use implies about the value of learning. When students believe artificial intelligence makes tasks pointless, their more self-driven forms of motivation tend to drop and external-pressure motivation tends to rise.

Controlled motivation means engaging because of pressure, demands, or avoidance of negative consequences, not because the activity feels meaningful. In practice, that can look like using artificial intelligence to comply, keep up, or reduce risk, rather than to deepen competence.

Where it fits

The pattern aligns with Self-Determination Theory, which distinguishes autonomous motivation (intrinsic enjoyment and identified personal value) from controlled motivation (pressure and obligation). The study did not test outcomes like grades, but it clarifies measurable pathways by which artificial intelligence could change the “why” behind student effort.

Notably, measurement invariance tests indicated the scale functioned similarly across gender, study level, academic field, and frequency of artificial intelligence use, enabling fairer comparisons between groups.

How to use it

Use the scale to separate “artificial intelligence as support” from “artificial intelligence as replacement.” If identified motivation is high (“support, not replace, my learning”) but redundancy beliefs rise, redesign tasks to reward thinking processes, not just outputs.

Watch for controlled-motivation cues, including perceived pressure to use artificial intelligence. Set clear norms that protect autonomy, such as allowing multiple tool pathways and explicitly valuing original reasoning, drafts, and reflection.

Limits & what we still don’t know

Several constructs were assessed with very short subscales, including two-item intrinsic and controlled motivation measures with modest reliability. Autonomy and competence in the artificial intelligence context were assessed with single items, which can miss nuance.

The amotivation indicator did not fit well in the confirmatory model and was treated as descriptive only. Differential item functioning appeared for some items across fields, meaning certain groups endorsed specific items differently than expected from their underlying trait levels.

Closing takeaway

This study’s contribution is measurement: a workable way to track whether artificial intelligence adoption is paired with meaning and choice, or with redundancy and pressure. If heavier use coincides with “this is pointless” beliefs, motivation problems may be predictable, not mysterious. The practical win is early detection and targeted course design before disengagement becomes the default.

Data in this article is provided by PLOS.

Related Articles

• Technology-enhanced neuromotor rehabilitation improved autonomy and well-being more than standard training in stroke and osteoarthritis
• Stress and coping sit upstream of multiple modifiable Alzheimer’s disease risks in network models
• Girls report higher body dissatisfaction than boys across countries, and links to lower well-being are stronger
• Performance crises in professional soccer grow from hidden vulnerabilities and escalating cycles, coaches report
• Virtual therapist tool was acceptable to hospitalized people with schizophrenia, but empathy and flexibility were limited
• Mixed physical and mental fatigue cut badminton smash speed 10.6 percent and accuracy 46.1 percent
• Virtual reality did not significantly change rowing muscle fatigue in trained men during ergometer exercise
• Nasal temperature drops during stress, especially social speech stress
• High-flow nasal therapy costs more than low-flow oxygen in COAST
• LIVEBORN newborn resuscitation feedback proved feasible and usable
• Complex chronic illness in older sexual and gender minority adults tied to higher Medicare spending and use
• Indian adolescents spent almost nine hours a day sedentary, with private school students sitting much more
• Nurse-delivered brief counselling reduced anxiety after self-poisoning at six months, but not at one year
• Patient reported frailty score after stroke predicted higher mortality in a Swedish registry study
• On-site and mixed Tai Chi cut anxiety and depression in college students over eight weeks
• Pharmacy teams accepted a mental health inpatient risk tool and refined it after early usability feedback