An emerging study reveals a curious paradox about the incorporation of artificial intelligence (AI) into the workplace: almost 50% of employees who interact with AI claim that its outputs cannot be trusted. This seems to be the case even as most people seem to be using AI quite regularly. An interdisciplinary team from Australia’s University of Melbourne and international consulting company KPMG AI studied this issue. They collected trust, use, and attitude data relating to AI from over 48,000 participants spanning 47 countries from November 2024 to January 2025.
Deficit of Trust in AI Systems
The research indicated that more than two-thirds of the respondents reported using AI for work, school, or personal activities, but only 46 percent of this group was willing to place their trust in these systems. In the survey, participants were asked to evaluate trust with regard to AI and technical capabilities along with the safety, security, and ethical soundness of the system, rating each on a nine-point scale to measure personal trust accurately.
With regard to AI, the authors of the study believe the components most people find hardest to believe in are the systems’ purported capabilities to be fair and to defend against harms. Contrary to this, where respondents reportedly claim to have less trust is in the technical capability of AI to accurately and reliably provide outputs and services.
Employing AI Technology in Business Environments
This research singularly analyzed the application of AI technology in professional environments. From this study, it was discovered that 58 percent of all respondents admit to using AI at the workplace frequently. A considerable subset of this population, 33 percent of all survey respondents, indicated that they actively use AI tools at work on a weekly or daily basis.
Those employees who reported using AI at the workplace claimed to gain benefits that include improved efficiency, enhanced access to information, and the ability to perform innovation in their work tasks. AI was claimed to be used in almost half of the cases reported by the respondents in a way that, in their estimate, activity aimed at generating revenue had been improved relative to such activity.
Risks and Policy Breaches
The report, however, emphasizes risks stemming from the rampant use of AI technology in the workplace. It shows that fifty percent of respondents who interacted with chatbots during work hours admitted doing so in violation of company policies. KPMG’s managing director, Samantha Gloede, shed some light on the unwarranted pressure employees seem to endure in regard to the use of AI technologies.
As Gloede put it, “People feel almost pressured that if they don’t use it, they will be… set behind their competitors.” This implies some degree of willingness to circumvent rules simply to compete. In her remarks, Gloede also noted how they have heard about employees engaging in dangerous activities like feeding sensitive company data into publicly available tools or forging deepfake videos of top executives. Instead of helping, these activities could tarnish the reputation of the individuals and the organization.
Presenting AI Work as One’s Own
Another risk identified is that of an employee using the output from AI chatbots and passing it off as their own origin. The research revealed, shockingly, that 57% of those surveyed admitted to concealing AI usage in their work. Most concerning, these individuals, as reported, did so without in any way gauging the rigor and validity of the content the AI produced for them. This poses questions about the trustworthiness and fidelity of work done with AI tools.
Being Misled because of Lack of AI Verification
The report goes on to note that a significant proportion of employees are being misled by AI and this uncorroborated reliance is causing mistakes. Per the study’s findings, 56% of respondents acknowledged the presence of errors in their work, which was a direct consequence of using AI unchecked. This exemplifies the trust gap combined with the disregard for thorough verification that accompanies the use of AI tools in the workplace.
Lack of AI Literacy and Training
The report highlights a critical gap in the understanding of the technology and the training AI employees obtain. For example, approximately 50 percent of the participants in this study reported that they did not comprehend AI and its applications. Furthermore, only 40 percent of the employees reported receiving any training or education specific to the appropriate and responsible AI usage in the workplace. Gloede was quoted as stating, “For organizations that don’t have an AI literacy study in place… [employees] are taking things into their own hands,” which suggests that employees are inappropriately defaulting to unsafe practices without the proper structure.
Recommendations and Future Gains
Samantha Gloede, managing director at KPMG, suggested that companies could introduce a “trusted AI framework” to train employees on the use of technology in their work. The framework could contain 10 guidelines for employees to learn about technology usage. Gloede hoped that C-suite executives, techs, and public policymakers would act on the survey findings. Trust is a strong predictor of AI’s acceptance, and organizations can move faster than the speed of trust. AI’s opportunity lies in responsible execution by organizations and governments, but there is a trust gap and lack of clear policies. The recommendations emphasize the urgent need for AI-expert education, clear policies, trust, and guidance.
