Workers see AI as helpful, but fear losing credit for their own expertise

Artificial intelligence (AI) could reduce routine tasks, ease workloads and improve working conditions, but workers may not experience those gains as better jobs if the same technology also weakens recognition, autonomy, social connection and professional identity, states a new study published in Proceedings of the ACM on Human-Computer Interaction.

The study, titled AI in the Workplace: The Impact of AI on Perceived Job Decency and Meaningfulness, finds that AI’s impact on job satisfaction varies sharply by sector and depends on whether the technology protects both job decency and job meaningfulness. The research is based on interviews with 24 workers in information technology, healthcare and service roles.

AI could improve decent work, but not equally across sectors

The study challenges a dominant workplace AI narrative that measures the technology mainly through productivity, efficiency and automation gains. The authors found that employees judge AI not only by whether it helps them complete tasks faster, but also by whether it protects the human value of work.

Job decency, as the study defines, is the basic conditions that make work fair and sustainable. These include working hours, work-life balance, job security, growth opportunities, compensation and workplace conditions. Job meaningfulness covers the deeper experience of work, including purpose, autonomy, challenge, social connection, recognition, pride and professional identity.

Participants described their current working lives and then imagined how future AI systems might affect their roles. The findings show that AI’s effects are neither uniform nor clearly positive or negative. IT workers generally expected AI to improve both work conditions and parts of meaningful work. Healthcare workers expected improvements in working conditions but more limited gains in meaningfulness. Service workers expected little improvement in basic work conditions but saw possible gains in social status from working with advanced technology.

Participants in the IT sector imagined AI as a workplace assistant that could summarize meetings, highlight action items, support scheduling, suggest code, prepare discussion materials and help with technical problem-solving. Many expected AI to reduce repetitive work and open more time for complex tasks. They also expected stronger growth opportunities as workers learn to use AI tools and adapt to new technical demands.

For IT workers, AI was often seen as a support system that could raise the quality of work rather than simply replace tasks. Participants expected improvements in work-life balance, working conditions and skill development. Some also expected AI to increase their professional value if it helped them solve harder problems and perform at a higher level.

Similarly, healthcare workers also expected AI to improve several dimensions of job decency. Participants imagined AI systems that could retrieve patient histories, monitor patient status, detect abnormalities in reports and scans, support diagnoses, assist with documentation, compare scans and flag overlooked clinical details. These tools were expected to reduce workload, lower stress and improve access to information.

The potential benefits in healthcare were closely tied to the pressure already placed on medical professionals. Participants described long hours, poor work-life balance and heavy responsibility as common features of the profession. AI was expected to relieve some pressure by handling documentation, monitoring and information retrieval, even if clinical responsibility remained with doctors.

Service workers were more skeptical about AI’s ability to improve basic working conditions. Participants imagined AI systems that could take orders, serve food, manage inventory, clean, estimate customer volume and generate operational reports, but many did not expect AI to improve working hours or work-life balance because customer demand, shift schedules and the nature of service work would remain largely unchanged.

Some service workers also anticipated that AI could add new burdens. If AI systems make mistakes, employees may have to supervise, correct or compensate for those failures. In that scenario, AI would not simply reduce work; it could create another layer of responsibility.

Across all three sectors, job security concerns were present but not dominant. Many participants believed human workers would remain necessary because accountability, judgment, customer care and patient responsibility cannot be fully delegated to AI. Still, some workers expected job security to decline as AI systems become more capable, particularly in roles where tasks are viewed as repetitive or easily automated.

In terms of compensation, IT workers expected skill growth with AI could improve their professional value. Service workers expected little change because pay was seen as shaped by labor protections, tips or existing wage structures. Healthcare workers raised concerns that organizations might offset the cost of AI adoption by reducing human compensation or redirecting patients toward AI-enabled systems.

Meaningful work faces a different risk

AI may improve job decency without protecting job meaningfulness. A job can become easier, faster or less stressful while also becoming less fulfilling, less socially valued or less connected to human identity. IT workers generally expected AI to support meaningful work by giving them more room for complex problem-solving, professional growth and autonomy. They imagined AI as a tool that could handle routine tasks while humans continue to guide goals, decisions and final outcomes.

However, even IT participants identified a risk: AI could make human expertise less visible. If colleagues or the public assume AI is doing most of the work, recognition for human judgment, problem-solving and effort may decline. This could weaken satisfaction even when the worker remains essential to the outcome.

Healthcare workers expected their jobs to remain meaningful because patient care, human responsibility and clinical judgment would still matter. Many saw AI as a tool that could assist diagnosis and monitoring but not replace the moral and professional responsibility of treating people. However, they also identified a risk to professional image and recognition. If patients or colleagues begin to see AI as the primary source of diagnosis, memory or treatment guidance, medical expertise could appear less distinctive. The study suggests that AI may improve clinical support while weakening visible recognition for the doctor’s skill.

Service workers saw a different trade-off. Human interaction is a major source of meaning in service jobs. Daily contact with customers and coworkers can make otherwise routine work feel socially valuable. If AI reduces the number of human coworkers or takes over customer-facing interactions, service workers may lose part of what makes the job meaningful. They also expected AI could raise their social image. Some participants believed that being able to work with advanced technology could make service roles appear more skilled, modern and respected. In this sector, AI could weaken some sources of meaning while strengthening others.

The study highlights a major workplace design flaw: human labor can become less visible when AI handles the most noticeable parts of a task. Workers may still supervise systems, correct errors, make final decisions and carry accountability, but those contributions may be hidden behind AI-generated outputs.

Monitoring AI systems, interpreting results, correcting mistakes and learning to use new tools require skill and judgment. If employers reward the technology while overlooking the human oversight that makes it useful, AI could reduce job satisfaction even when it improves productivity.

Workplace relationships are another key concern. Healthcare workers emphasized teamwork in patient care. Service workers valued interaction with colleagues and customers. IT workers valued mentoring, informal conversations and collaborative problem-solving. AI systems that reduce human contact too aggressively could damage social connection across sectors.

The findings suggest that AI does not merely change task allocation. It changes how work is interpreted, credited and valued, making workplace AI not only a technical or economic issue, but also a social and organizational issue.

Employers need AI designs that protect human value

Organizations should move away from generic workplace AI strategies, the study insists. A system that improves job satisfaction in one sector may reduce it in another if it disrupts the values that workers attach to their jobs.

• For IT workers, AI should support learning, autonomy and complex problem-solving while keeping human expertise visible. AI tools should not make workers appear replaceable or reduce recognition for the judgment behind technical work.
• For healthcare workers, AI should support clinical decision-making without undermining professional authority or patient trust. Doctors and other health professionals need to remain clearly accountable and visibly central to care, especially when AI assists with diagnosis, documentation or monitoring.
• For service workers, AI should reduce physical strain and repetitive work without eliminating the customer and coworker interactions that give the role social value. AI should make service workers appear more capable, not less necessary.

The authors identify several design priorities. AI systems should preserve workers’ social image, acknowledge unseen human effort, strengthen human-human interaction, balance autonomy and control, and support career growth. These priorities go beyond efficiency and address the conditions that make work satisfying over time.

Autonomy is particularly important. In decision-heavy roles such as healthcare, AI recommendations could feel threatening if workers cannot challenge or adjust them. In service work, however, AI could increase autonomy if employees feel they are directing AI tools rather than being controlled by them. The same design choice can therefore carry different meanings in different sectors. The study also warns against assuming that removing repetitive tasks always improves work. In some jobs, handing routine work to AI can free people for more meaningful challenges. In others, it can leave workers with only stressful, emotionally heavy or less varied responsibilities.

The future of work, the study suggests, is not just about whether AI will replace jobs. It is also about how AI will change the quality of work that remains. AI may redistribute tasks, alter expectations, change recognition patterns and reshape how skills are valued.

Implications and limitations

AI should not be introduced only as a productivity tool; its success should also be judged by whether it preserves decent and meaningful work. This means reducing workload without weakening autonomy, improving efficiency without hiding human contribution, and supporting automation without eroding recognition, trust or workplace relationships.

For companies adopting AI, the next challenge is not simply to automate tasks or improve output. It is to design systems that protect both job decency and job meaningfulness. In real-world settings, this requires making human oversight visible, recognizing the labor involved in correcting and supervising AI systems, preserving worker control over AI recommendations, and ensuring that employees continue to receive credit for expertise, judgment and accountability.

The findings also suggest that workplace AI strategies must be sector-specific. IT workers may benefit from AI that supports learning and complex problem-solving. Healthcare workers need systems that assist clinical judgment without weakening professional authority or patient trust. Service workers may gain from AI that reduces physical strain, but only if it does not remove the customer and coworker interactions that give the work social value.

The researchers acknowledge several limitations. The findings are based on workers’ expectations rather than long-term observation of AI deployment. The study involved a relatively small sample of 24 participants, with more men than women and no non-binary participants. Regional differences in labor laws, workplace cultures and AI adoption practices may also affect how workers experience AI in real workplaces.