AI at Work: Trust and Clear Rules Matter More Than Hype

Artificial intelligence is rapidly entering workplaces, but a new Kuwait-based study suggests that adoption will depend less on excitement around new tools and more on whether organizations can build trust, provide clear policies, and support employees through change.

The study, "To AI or Not to AI? When Synergies Collide," by Sapheya Aftimos of Australian University and Randa Diab-Bahman of American International University, was published in MDPI Proceedings and presented at the International Conference on Digital Transformation, Sustainability and AI in Kuwait.

Based on survey responses from 153 participants across education, oil and gas, telecommunications, and banking, the research finds that AI usage is strongly linked to two factors: positive stakeholder sentiment and supportive institutional policies.

Beyond the AI buzz

Across industries, employees are increasingly encouraged to use AI to improve productivity, speed up analysis, automate routine tasks, and support decision-making. However, many workplaces still lack clear rules on what AI tools can be used, what data can be shared, when human oversight is required, and who is accountable when AI-supported decisions go wrong, creating a confusing environment.

Workers may be told to innovate with AI, but they may also fear breaching confidentiality, producing inaccurate work, or violating unclear institutional rules. Managers may want efficiency gains, but lack governance frameworks to guide implementation. Institutions may promote digital transformation, while leaving employees uncertain about risks and responsibilities.

The study examines this tension through a three-part framework: institutional policies, stakeholder sentiments, and AI usage. Rather than treating AI adoption as a standalone technical issue, the researchers argue that usage is shaped by the interaction between organizational rules, human attitudes, and workplace capability.

Trust drives adoption

The study found a strong relationship between stakeholder sentiment and AI usage. The researchers report a Pearson correlation value of 0.80, showing a strong positive association between favorable attitudes toward AI and reported AI use. In simple terms, people are more likely to use AI when they believe it is useful, trustworthy, and relevant to their work. If they see AI as unreliable, risky, threatening, or poorly explained, they are less likely to adopt it meaningfully.

Many organizations assume that adoption will follow once tools are made available. The study suggests otherwise. Access alone is not enough. Employees need confidence. They need to understand how AI works, where it helps, where it fails, and how their organization expects them to use it.

In sectors such as banking, education, telecommunications, and oil and gas, which handle sensitive data, safety concerns, public interest responsibilities, and high-value decisions, trust is crucial. AI use in these settings cannot be treated as casual experimentation. It requires confidence in both the technology and the institution managing it.

Clear policies can unlock responsible innovation

The study also finds a strong positive relationship between institutional policies and AI usage, with a Pearson correlation value of 0.73. This suggests that stronger institutional support, leadership backing, and resource allocation are associated with higher AI adoption.

This finding challenges the idea that policy always slows innovation. In the case of AI, clear rules may actually encourage responsible use. When employees know what is allowed, what is prohibited, and what safeguards apply, they are more likely to use AI with confidence. Weak or absent policies can have the opposite effect. Some employees may avoid AI because they fear making mistakes. Others may use public or unapproved tools informally, creating risks around data privacy, intellectual property, compliance, and accountability.

AI governance should not be limited to broad ethical statements in organizations. Employees need usable guidance: approved tools, data-sharing rules, human review requirements, documentation standards, and escalation routes when risks arise.

For governments, the findings suggest that national AI strategies must be translated into sector-level and workplace-level guidance. Ambition without implementation rules leaves institutions exposed.

Policies also shape how people feel about AI

The research reports a moderate positive correlation of 0.61 between institutional policies and stakeholder sentiment. In other words, better policies are linked with more positive attitudes toward AI. Governance does not only control behavior, it can also build confidence. When institutions communicate clearly, invest in training, and address ethical and operational concerns, stakeholders are more likely to view AI as manageable rather than threatening.

Fear and uncertainty often grow in policy vacuums. Employees may worry that AI will replace jobs, monitor performance, reduce autonomy, or expose them to blame for system errors. Students and educators may worry about academic integrity and learning quality. Customers may worry about automated decisions and data misuse.

Good policy cannot remove every concern, but it can create a more stable environment. It can show that AI adoption is being handled deliberately, not rushed blindly.

The wider lesson for fast-digitizing economies

Although the study is based in Kuwait, its relevance extends far beyond the country. Gulf economies are investing heavily in digital transformation, AI-enabled public services, smart infrastructure, and economic diversification. Additionally, many developing and emerging economies face a similar challenge: how to adopt AI quickly without weakening accountability, trust, or institutional control.

The study adds value because AI adoption research is often dominated by Western contexts or technology-producing economies. Evidence from Kuwait helps broaden the discussion and highlights the need for context-sensitive adoption models. For Global South stakeholders, AI readiness is not only about infrastructure, cloud systems, or access to digital tools. It also depends on human capital, governance capacity, public confidence, and institutional maturity.

Countries and organizations that invest in AI without investing in these foundations may see uneven adoption, informal tool use, workforce resistance, or public distrust. Those that align technology with policy and people are more likely to capture long-term benefits.

A development and governance issue

AI can support economic productivity, better education systems, improved industrial efficiency, stronger financial services, and smarter infrastructure, if implemented responsibly. The study is relevant to several Sustainable Development Goals, including SDG 4 on quality education, SDG 8 on decent work and economic growth, SDG 9 on industry, innovation and infrastructure, and SDG 16 on effective institutions.

• Businesses: Leaders need to move from informal encouragement to structured adoption. That means setting clear internal rules, training staff, explaining risks, monitoring use, and ensuring that AI supports rather than weakens human judgment.
• Policymakers: The study points to the need for sector-sensitive AI governance. Banking, education, energy, and telecommunications do not face the same risks. Regulation and guidance must reflect these differences.
• Development agencies and international organizations: The research reinforces the need to treat AI capacity-building as more than technical training. Institutions also need help building policy frameworks, ethical review systems, accountability mechanisms, and workforce readiness.
• Civil society: The study highlights the importance of oversight. As AI enters sectors that affect learning, finance, employment, energy, and public services, citizens need assurance that adoption will not undermine fairness, rights, safety, or transparency.

The evidence is useful, but not the final word

The sample size of 153 participants provides a focused view, not a complete national picture. The study is limited to Kuwait and does not show whether the same relationships would hold in countries with different regulatory systems, labor markets, or organizational cultures.

The findings are also based on self-reported survey data, which can reflect perceptions as much as actual behavior. The analysis identifies correlation, not causation. It shows that policies, sentiment, and AI usage are connected, but it does not prove that one directly causes another. The paper also does not provide detailed sector-by-sector results in the published text, leaving open important questions about whether AI adoption behaves differently in education, banking, telecommunications, and oil and gas.

The limits do not reduce the study's value, but point to where the next wave of research should go.

The next frontier: proving AI improves outcomes

Future research should test the same framework across more countries, specifically the Gulf, the wider Middle East, Africa, South Asia, and other Global South regions. Comparative studies could show how governance systems, labor markets, and institutional cultures shape AI adoption.

Sector-specific studies are also needed. AI in banking raises different concerns from AI in education or energy. Each sector needs its own evidence base on risk, trust, productivity, and accountability.

Long-term studies would be especially useful to demonstrate whether better policies actually improve trust over time, and whether stronger trust leads to sustained and responsible AI use. Interviews and case studies could also help explain why employees embrace or resist AI in real workplace settings.

Most importantly, future research should examine outcomes. More AI use is not automatically better. The key question is whether AI improves productivity, learning, safety, service quality, and decision-making without creating new harms.

To sum up, AI adoption will fail if institutions rely on hype while leaving users uncertain. It will succeed when policies are clear, employees are supported, and trust is built into the adoption process from the beginning.