New Research Explores How AI can Build Trust in Knowledge Work

In today’s economy, many workers have transitioned from manual labor toward knowledge work, a move driven primarily by technological advances, and workers in this domain face challenges around managing non-routine work, which is inherently uncertain. Automated interventions can help workers understand their work and boost performance and trust. In a new study, researchers explored how artificial intelligence (AI) can enhance performance and trust in knowledge work environments. They found that when AI systems provided feedback in real-time, performance and trust increased.

The study, by researchers at Carnegie Mellon University, is published in Computers in Human Behavior. The article is part of a special issue, "The Social Bridge: An Interdisciplinary View on Trust in Technology," in which researchers from a range of disciplines explore mechanisms and functions of trust in people and technologies.

"Our findings challenge traditional concerns that AI-driven management fosters distrust and demonstrate a path by which AI complements human work by providing greater transparency and alignment with workers’ expectations," suggests Anita Williams Woolley, Professor of Organizational Behavior at Carnegie Mellon’s Tepper School of Business, who co-authored the study. "The results have broad implications for AI-powered performance management in industries increasingly reliant on digital and algorithmic work environments."

Applications of machine learning and AI have consistently proven capable of performing demanding cognitive tasks, provided they can be routinized. But in non-routine work, AI capabilities (e.g., those designed to facilitate managers’ ability to monitor productivity) often backfire, fostering enmity instead of efficiency.

In this study, researchers sought to determine how the frequency of feedback and the uncertainty of a task interacted to influence workers’ perceptions of an algorithm’s trustworthiness. In a randomized, controlled experiment, 140 men and women (primarily White and with a median age of 39) performed caregiving tasks in an online, simulated home healthcare environment.

Individuals were randomly assigned to receive or not receive automated real-time feedback (i.e., feedback delivered during the task) while performing their work under conditions of high or low uncertainty. After completing the task, they received an algorithmically determined rating based on their actual performance on the task.

Real-time feedback increased the perceived trustworthiness of the performance rating by boosting workers’ sense of their own work quality (i.e., knowledge of the results) and reducing the degree to which they were surprised by their final evaluation. This, in turn, enhanced workers’ trust in AI-generated performance ratings - particularly in non-routine work settings where uncertainty was high.

Among the study's limitations, the authors note that their findings may not generalize to all circumstances, in part because study participants were not drawn from a population of caregivers and the simulated task did not represent actual caregiving. In addition, the study did not examine the role of individual differences, such as levels of conscientiousness and expertise.

"Non-routine work has long posed challenges to traditional management strategies, and the development of algorithmic management systems offers an opportunity to begin to address them," notes Allen S. Brown, a PhD student in Organizational Behavior and Theory at Carnegie Mellon's Tepper School of Business, who led the study. "Our identification of a new framework for examining managerial interventions, one that makes performance standards more transparent and increases workers' knowledge of the results, is particularly relevant in today’s emerging work environments."

The study was funded by the AI-CARING Project of the U.S. National Science Foundation.

Allen S Brown, Christopher R Dishop, Andrew Kuznetsov, Ping-Ya Chao, Anita Williams Woolley.
Beyond efficiency: Trust, AI, and surprise in knowledge work environments.
Computers in Human Behavior, Volume 167, 2025. doi: 10.1016/j.chb.2025.108605

Most Popular Now

AI Model can Read ECGs to Identify Femal…

A new AI model can flag female patients who are at higher risk of heart disease based on an electrocardiogram (ECG). The researchers say the algorithm, designed specifically for female patients...

New AI Tool Mimics Radiologist Gaze to R…

Artificial intelligence (AI) can scan a chest X-ray and diagnose if an abnormality is fluid in the lungs, an enlarged heart or cancer. But being right is not enough, said...

To be Happier, Take a Vacation... from Y…

Today, nearly every American - 91% - owns a cellphone that can access the internet, according to the Pew Research Center. In 2011, only about one-third did. Another study finds...

Researchers Find Telemedicine may Help R…

Low-value care - medical tests and procedures that provide little to no benefit to patients - contributes to excess medical spending and both direct and cascading harms to patients. A...

AI Revolutionizes Glaucoma Care

Imagine walking into a supermarket, train station, or shopping mall and having your eyes screened for glaucoma within seconds - no appointment needed. With the AI-based Glaucoma Screening (AI-GS) network...

North Cumbria Integrated Care Signs 10-Y…

North Cumbria Integrated Care NHS Foundation Trust (NCIC) has signed a long-term agreement for use of the Alcidion Miya Precision platform, to provide an electronic patient record (EPR) for the...

AI Accelerates Discovery of Neurodevelop…

Researchers have developed an artificial intelligence (AI) approach that accelerates the identification of genes that contribute to neurodevelopmental conditions such as autism spectrum disorder, epilepsy and developmental delay. This new...

AI may Help Clinicians Personalize Treat…

Individuals with generalized anxiety disorder (GAD), a condition characterized by daily excessive worry lasting at least six months, have a high relapse rate even after receiving treatment. Artificial intelligence (AI)...

AI can Open Up Beds in the ICU

At the height of the COVID-19 pandemic, hospitals frequently ran short of beds in intensive care units. But even earlier, ICUs faced challenges in keeping beds available. With an aging...

AI Model Predicting Two-Year Risk of Com…

AFib (short for atrial fibrillation), a common heart rhythm disorder in adults, can have disastrous consequences including life-threatening blood clots and stroke if left undetected or untreated. A new study...

Accelerating NHS Digital Maturity: Paper…

Digitised clinical noting at South Tees Hospitals NHS Foundation Trust is creating efficiencies for busy doctors and nurses. The trust’s CCIO Dr Andrew Adair, deputy CCIO Dr John Greenaway, and...

Mobile App Tracking Blood Pressure Helps…

The AHOMKA platform, an innovative mobile app for patient-to-provider communication that developed through a collaboration between the School of Engineering and leading medical institutions in Ghana, has yielded positive results...