Trust Is Earned. Here Are the Eight Principles That Build It.

Trust in artificial intelligence does not arrive automatically. It is built, tested, and maintained, or it erodes. Twenty years in technology and design makes that clear.

Organisations are adopting AI-powered systems at a pace that has outrun the frameworks needed to understand whether those systems deserve to be trusted. That gap matters. It is why this research began.

David Horsager's work in The Trust Edge defines eight components of trust. They were developed for human leadership, but they apply with equal force to the AI systems organisations now rely on every day.

‍

Why Trust in AI Cannot Be Assumed

Deploying an AI system does not make it trustworthy. A system can be accurate, fast, and capable, and still fail the people using it if it behaves inconsistently, cannot explain itself, or was not designed with the end user in mind. The eight principles below are the test.
‍

Clarity

People trust what they understand. Organisations and leaders who communicate with honesty and transparency earn trust over time. The same principle holds for AI. A system that cannot explain its decisions, or that operates as a black box, will never be fully trusted, regardless of how accurate it is.

‍

Compassion

Trust grows when people feel heard and considered. For leaders, this means active listening and genuine care for the people around them. For AI, it means systems designed with the end user in mind, not just the output they produce. Who is using this? What do they need?

‍

Character

Character is consistency of behaviour when no one is watching. For a person, it is ethical decision-making under pressure. For an AI system, it is behaving predictably and safely even in edge cases, even when the prompt is unusual, even when the stakes are low. Character cannot be performed. It has to be built in.

‍

Competence

Competence is the willingness to keep learning, to acknowledge what you do not know, and to improve. No leader, and no AI system, is trusted for being infallible. They are trusted for being capable and honest about their limitations. Overconfidence is one of the fastest ways to lose trust in both contexts.

‍

Commitment

People follow leaders who stay the course when things are hard. The same dynamic applies to technology. Systems that are maintained, updated, and supported over time earn the trust of the organisations using them. Abandonment, whether of a team or a platform, ends trust quickly.

‍

Connection

Trust is relational. Leaders build it through authenticity, through being present, through avoiding behaviour that undermines others. For AI systems, connection is about the relationship between the system and the people who use it. Does it feel designed for them? Does it respect their time?

‍

Contribution

Contribution is about showing up. Investing effort toward shared goals, not just claiming credit for outcomes. An AI system earns trust by genuinely contributing, by saving time, reducing friction, and improving results, not by adding complexity in pursuit of appearing capable.

‍

Consistency

This one is non-negotiable. Consistency is the thread that holds every other element together. A single act of inconsistency can undo months of trust-building. For AI systems, this is the highest bar. Inconsistent output or inconsistent behaviour destroys confidence faster than any other failure mode.

‍

How These Principles Shape the Way Envent Builds

Trust is earned. Not assumed, not declared, not granted by default. It is built through daily practice across all eight of these dimensions.

This framework is foundational to Justin Dean's PhD research at the University of Newcastle, examining how organisations can develop warranted trust in artificial intelligence. The question driving that research is not whether AI is capable. The question is whether it deserves to be trusted, and how organisations would know.

At Envent, it shapes how we think about every system we design and every product we build. If a person cannot trust it, it is not finished.

‍

‍