People can use artificial intelligence every day without trusting the organisations deploying it. That contradiction is becoming a business risk as assistants and automated decisions move into customer service, education, internal workflows, and personal life.
KPMG research cited in the source found that 97 percent of people in the United Arab Emirates use AI for work, study, or personal purposes. Yet 84 percent would trust it only if they felt confident it was being used responsibly, and 57 percent wanted stronger rules.
The United Kingdom shows a similar gap. Only 42 percent were willing to trust AI, although 57 percent accepted its use. Eighty percent wanted stronger regulation, 78 percent worried about harmful outcomes, and just 10 percent said they understood the rules already in place. Low awareness means even existing safeguards may do little to reassure people if companies cannot explain them clearly.
Usage Does Not Equal Permission
Rapid adoption can tempt leaders to treat trust as settled. If employees use generative tools and customers interact with automated support, the technology appears socially accepted. The survey results suggest something narrower: people value the convenience while remaining uncertain about how decisions are made and who is accountable.
That distinction matters when automation represents a brand. An inconsistent customer-service response, an undisclosed bot, or a decision no employee can explain may damage confidence beyond the individual interaction. In a market where four out of five respondents want stronger controls, opacity is not merely a technical concern.
Lei Gao, chief technology officer at SleekFlow, argues that adoption is no longer the central problem; accountability is. Trust grows when AI behaves predictably and transparently and declines when users cannot tell whether automation or a person is deciding.
Transparency and Human Control Must Be Designed In
The first practical step is disclosure. Customers and employees should know when they are interacting with AI and when a human has taken over. Clear handoffs prevent a system from borrowing human authority while avoiding human responsibility. They also give users a realistic expectation of what the interaction can resolve.
AI should also extend people rather than quietly remove them from consequential workflows. Internal adoption is easier when staff understand how a tool supports their judgement, what authority it possesses, and where they can intervene. Human control must be an operating mechanism, not a reassuring phrase on a policy page.
The same principles apply whether an organisation uses foundation models through AWS Bedrock, information infrastructure in a Dell AI Factory, or assistants such as SAP Joule. Trust depends less on the vendor’s label than on how the company configures, communicates, and supervises the deployment.
Responsible Operation Continues After Launch
AI behaviour requires ongoing review for tone, fairness, bias, consistency, and escalation quality. A model or connected information source can change, customer language can expose new edge cases, and a workflow that seemed low-risk may acquire broader authority over time.
Monitoring should therefore connect technical measures with human outcomes. Accuracy alone cannot reveal whether a response was needlessly harsh, whether one group receives poorer service, or whether employees feel pressured to accept an automated recommendation. Complaint patterns and escalation quality can reveal trust failures that aggregate performance dashboards miss.
The UAE demonstrates that use can spread faster than trust. The UK figures show that familiarity does not automatically close the gap. For leaders, success can no longer be counted only through interactions, licences, or hours saved.
The next credible milestone is proof that automation serves people consistently. That requires visible disclosure, reliable human escalation, and evidence that the system is reviewed after deployment. Businesses will earn trust not by insisting that AI is trustworthy, but by making accountability observable whenever the technology acts.
