Using ChatGPT at work? Your boss may trust you less, new study finds
image:
Organizational Policy Affects Trust Penalty for Using LLMs
view moreCredit: (Credit- Yochanan Bigman)
A new study reveals a hidden cost of using artificial intelligence at work: employees who rely on tools like ChatGPT are trusted less by their supervisors, consequently harming their performance evaluations and promotion prospects. Across six studies, researchers found that managers consistently penalized AI users, even when AI use was not explicitly prohibited. Organizations that actively encouraged AI use substantially reduced this trust penalty, highlighting the critical role of workplace policies in shaping the future of human-AI collaboration.
Artificial intelligence may be making employees faster and more productive, but it could also be quietly damaging their careers. A new study led by Samuel Meyers, a Ph.D. student, together with his advisor Dr. Yochanan Bigman of the Hebrew University of Jerusalem and Dr. Roman Briker of WHU – Otto Beisheim School of Management in Germany, suggests that workers who use ChatGPT and other AI tools may pay an unexpected professional price: their managers are less likely to trust them.
The research, published in the prestigious scientific journal The Leadership Quarterly, reveals that supervisors consistently place less trust in employees who rely on artificial intelligence to complete their work, even when using such technology is not explicitly prohibited.
Across six studies, the researchers examined how employees' use of large language models (LLMs), such as ChatGPT, influences their supervisors' trust and subsequent professional evaluations.
The hidden price of using AI
In one experiment, participants acting as supervisors were asked to decide whether to trust an employee's report about time spent completing a task or verify it themselves. The supervisors knew whether the employee had produced the work independently or used AI.
The results were striking: employees who completed the work themselves had a 68% chance of earning their supervisors’ trust, compared with just 24% when they used AI. Importantly, supervisors were not evaluating the quality of the finished work. Their decisions were based on knowing how it had been produced, suggesting that the use of AI itself can influence judgments about an employee's trustworthiness.
The researchers also compared AI-assisted work with work copied from another person. While using AI reduced trust compared with independent work, it was still viewed more favourably than plagiarism.
The implications could extend far beyond everyday workplace relationships. Additional experiments showed that lower supervisory trust was associated with poorer evaluations of employee performance, reduced perceptions of employees' contributions to the organization, and lower perceived suitability for promotion.
“AI is changing not only how we work, but also how we judge the people we work with,” said Dr. Bigman. “An employee may use AI to produce better work more efficiently yet still lose their manager's trust simply because AI was involved. Organizations need to recognize that adopting AI is not just a technological challenge — it is also a psychological and managerial one.”
Can company policy change the picture?
Perhaps the most encouraging finding concerns the role of employers themselves.
The researchers discovered that when organizations explicitly encouraged employees to use AI, the negative effect on supervisory trust was substantially reduced and was no longer statistically significant.
In contrast, when AI use was prohibited — or when no organizational policy existed — supervisors were significantly less likely to trust employees who used it.
Surprisingly, neither the importance of the task nor supervisors' own experience with AI consistently eliminated the trust penalty.
The findings were further supported by a study involving academic supervisors evaluating real work by graduate students. The more extensively students used AI, the less their supervisors tended to trust them.
A wake-up call for employers
As AI becomes increasingly integrated into workplaces, the researchers warn that organizations may face a growing dilemma: employees are encouraged to embrace technologies that improve productivity, while simultaneously risking negative judgments for using them.
This tension could even encourage employees to conceal their AI use, creating what researchers call “shadow AI” — the undisclosed use of artificial intelligence at work.
The findings suggest that employers should go beyond simply providing access to AI tools. Clear organizational policies that actively support responsible AI use, alongside appropriate managerial training, may help protect workplace trust.
For employees, the message is equally important: the benefits of AI may depend not only on what the technology can accomplish, but also on how managers perceive those who use it.
The study highlights a paradox of the AI revolution: the technology designed to make us better at our jobs may also make our bosses trust us less.
Journal
The Leadership Quarterly
Method of Research
Survey
Subject of Research
People
Article Title
Psychological reactions to subordinates’ LLM use at work
Chatbots with personality?
Study by Mainz University shows how AI language models express and recognize individual personality traits in texts
Texts generated by so-called large language models, or AI-powered chatbots, exhibit characteristic “personality” traits. Such chatbots can also identify individual personality traits in texts. However, humans are more reliable at distinguishing between human-written and AI-generated texts. These are the findings of researchers at the Institute of Psychology at Johannes Gutenberg University Mainz (JGU), as reported in the journal Scientific Reports.
“Chatbots are becoming part of more and more areas of our lives, from customer service and digital counseling services to personal chatbots that many people already use every day and even share their feelings with,” said Sophie Körner, a research associate in the Department of Personality Psychology and Psychological Assessment at JGU and lead author of the study. “That is why we need to understand how we perceive chatbot-generated texts and whether we can still distinguish them from human-written texts.”
More conscientious, agreeable, and emotionally stable than humans
For their study, Körner, together with Dr. Sarah Osterholz and Professor Boris Egloff, compared the self-descriptions of seven AI language models – ChatGPT-4o, ChatGPT-3.5, Llama-3, Gemma 2, Gemini 2.5, Phi-3, and Orca 2 – with those of 70 humans. Using psychological questionnaires and open-ended questions, they asked the chatbots and human participants to describe their respective “personalities” in German. For example, they were asked to rate statements about their self-esteem and the so-called Big Five personality traits, such as openness, agreeableness, and extraversion. The subsequent open-ended questions focused on their interests, hopes for the future, and typical behavior in stressful situations. Each language model was questioned ten times, giving the researchers a total of 140 texts to analyze.
The team first examined how similar the language models' responses were when questioned repeatedly. “Our results show that the language models do indeed attribute fairly stable personality profiles to themselves over time,” said co-author Osterholz. Compared with the human participants, the language models presented themselves as more conscientious, agreeable, and emotionally stable.
Humans are more reliable at identifying AI-generated texts
Next, the researchers asked ten humans and six chatbots to evaluate the self-descriptions. Based on the texts, they were asked to assess personality traits such as openness, conscientiousness, and self-esteem. Both humans and chatbots were able to draw conclusions about personality traits. However, the chatbots tended to rate some traits of the human authors, including openness and self-esteem, higher than the human raters did. Conversely, the human raters tended to assign higher scores to the language models than to the human-written texts for all traits examined, particularly emotional stability.
The raters were also asked to determine whether each text had been written by a human or an AI model. The human raters achieved an accuracy of 83 percent, compared with just 67 percent for the chatbots. The two best-performing AI models in this task, Gemma and Gemini, were about as reliable as the average human rater at identifying whether a text was written by a human or AI. Both groups tended to classify texts as human-written more often, possibly to avoid misattributing human-written texts to AI. Specifically, chatbot-generated texts were more likely to be incorrectly classified as human-written than vice versa. However, AI models may become significantly more accurate at identifying whether a text was written by a human or AI as their training data increasingly include AI-generated texts.
The human raters primarily cited the length of AI-generated texts and their stilted, particularly technical or analytical language as distinguishing features. A computer-assisted linguistic analysis also revealed differences: AI-generated texts contained more time-related expressions and function words such as articles and prepositions, while human self-descriptions more often focused on leisure activities, feelings, and social relationships.
Potential applications in psychological assessment
According to the researchers, AI language models could also be used in psychological assessment and human resources in the future, for example, to obtain an initial picture of personality traits. Such applications are already being pursued in some workplace settings. However, it remains unclear how reliable these methods are and under what conditions they could be used. “They could provide an automated, easily accessible initial assessment,” said Körner. But since language models are trained on human-written texts, they also acquire biases, meaning distortions in judgment. It is therefore important to carefully examine whether personality assessments in sensitive areas should be left to algorithms, particularly from an ethical perspective.
Journal
Scientific Reports
Method of Research
Data/statistical analysis
Subject of Research
People
Article Title
Personality expression and perception in large language models and humans
Article Publication Date
3-Nov-2026
No comments:
Post a Comment