AI at Work Can Boost or Harm Performance Depending on Mindset

As companies race to integrate artificial intelligence into the workplace, new psychological research shows that the technology can either unlock higher productivity or quietly erode employee performance, depending largely on an individual’s mindset. The study, published in the journal Humanities and Social Sciences Communications, reveals that everyday AI usage simultaneously activates two opposing psychological mechanisms in workers, determining whether they feel empowered or overwhelmed.

The Double-Edged Nature of Workplace AI

Artificial intelligence is rapidly transforming day-to-day operations across various industries, but its psychological impact on staff has remained difficult to untangle. While enterprise leaders frequently promote AI as an unalloyed productivity driver, the reality inside organizations appears far more complex.

To investigate this phenomenon, researchers Wenhui Zhang, Po‐Chien Chang, and Xinqi Geng conducted a three-wave field study tracking 409 employee-supervisor pairs in AI-intensive industries across China. By collecting matched survey data from employees and their direct managers, the team discovered that AI usage operates as a classic double-edged sword, setting off helpful and harmful behavioral chains at the exact same time.

Two Competing Pathways: Empowerment vs. Pressure

Relying on the framework of self-determination theory, the study identified two parallel chains connecting AI tools to employee task outcomes. On the constructive side, AI can foster autonomous motivation, giving employees a sense of volition, ownership, and mastery over their roles. In response, workers engage in promotion-focused job crafting—actively expanding their job boundaries, seeking out novel challenges, and ultimately driving up their task performance.

Conversely, the study found that AI can just as easily generate controlled motivation. In this pathway, employees perceive the technology as an overbearing monitoring system or an algorithmic competitor, creating feelings of pressure and micromanagement. To shield themselves from criticism and errors, workers turn to prevention-focused job crafting. Rather than innovating, they adopt defensive, risk-averse behaviors, strictly shrinking their responsibilities to the bare minimum—a reaction that measurably degrades their overall performance.

The Protective Shield of High Self-Evaluation

Which of these two pathways wins out largely depends on the employee’s internal psychological resources. The researchers found that an individual’s core self-evaluations—their baseline assessment of personal competence, self-worth, and internal control—serves as a vital boundary condition.

According to the findings, workers with high core self-evaluations capitalized on AI tools as enabling partners, which amplified autonomous motivation and proactive behaviors. At the same time, this strong self-regard acted as a protective buffer, shielding them from the anxiety and defensive behaviors triggered by algorithmic scrutiny. In contrast, employees with lower self-evaluations were far more susceptible to the stress-inducing downsides of digital automation.

Rethinking How Companies Deploy AI

The findings indicate that simply rolling out advanced digital tools is insufficient to guarantee organizational gains. If companies introduce software without considering worker psychology, they risk driving employees toward protective disengagement and lower productivity.

To ensure AI serves as an asset rather than a liability, the researchers suggest that leaders adopt personalized integration strategies. Rather than treating automation as a pure efficiency metric, organizations should cultivate environments that support employee autonomy and actively bolster self-confidence, ensuring workers view artificial intelligence as a support tool rather than a digital overseer.

---