.svg)
Building Trust in AI Development: Connection
Trust lives in relationships. How the Connection principle shapes AI people actually use.
.png){kind=avatar}
Trust is not a feature. It is a relationship.
This sounds obvious until you look at how most AI systems are built. They are designed for accuracy, speed, and capability. Rarely are they designed with the question "how does this feel to the person using it?" as the primary design brief. That gap is where AI adoption fails, and where the Connection principle becomes one of the most important trust factors to get right.
What Connection Means in Horsager's Framework
David Horsager's eighth trust principle, Connection, is built on a straightforward premise: people trust those who show genuine care for them. Not systems. Not processes. People.
When applied to human leadership, Connection means making others feel seen, heard, and respected. It means expressing real appreciation, listening without agenda, and refusing to erode relationships through gossip or disengaged behaviour. Every interaction, Horsager argues, either deposits into or withdraws from the trust account. There is no neutral.
In AI development, the same logic holds. Every interaction a person has with an AI system either builds their confidence in it or erodes it. A response that misses the point of what someone asked, an interface that makes users feel stupid for not understanding it, a system that contradicts itself across sessions, a chatbot that responds to distress with a support article link. These are withdrawals. Small ones, maybe. But they accumulate.
The Difference Between Capable and Connected
An AI system can be technically accurate and still feel disconnected. Healthcare wayfinding is a good example. A patient who is anxious, running late, and navigating an unfamiliar campus does not need the shortest-path algorithm. They need a system that acknowledges their context. One that communicates in plain language. One that does not make them feel like they have failed if they need to ask the same question twice.
This is what Justin Dean's research at the University of Newcastle is investigating. How do people form trust relationships with AI systems? What makes a person feel that a system is on their side, not just processing their inputs? The answers are not purely technical. They are relational.
Envent's approach to digital wayfinding has always treated the user as a person in a specific emotional state, not a query to be resolved. Thirty years of field experience across hospitals, shopping centres, and government buildings has shown, repeatedly, that wayfinding systems fail not because the map is wrong, but because the interaction feels wrong. The connection is missing.
Building Connection Into AI Systems
Designing for connection in AI requires deliberate choices at every level of the system.
Language matters. A system that says "invalid input" instead of "I didn't quite catch that, can you try again?" is making a choice. Not a technical one. A relational one.
Tone matters. The difference between an AI that feels like a vending machine and one that feels like a knowledgeable guide is almost entirely in how it communicates, not in what it knows.
Memory and continuity matter. A system that treats every session as a clean slate forces users to re-establish context every time. That is tiring. It also signals that the system does not care about the relationship. Even basic session continuity, handled well, can dramatically change how connected a user feels.
Transparency matters. When an AI is honest about what it does not know, or routes a user to a human when it should, that is not a failure. That is a trust deposit. Users who feel that a system is being straight with them extend more goodwill when things go wrong.
Why This Matters for AI Adoption in Australia
Public trust in AI in Australia is still forming. It is not fixed. Which means the choices made now in how AI systems are designed to connect with people will shape how the next decade of AI adoption unfolds.
Organisations that build AI with Connection as a design principle, not an afterthought, will find that their systems are used more, trusted more, and generate better outcomes. Not because the technology is superior. Because the relationship is.
Envent builds wayfinding AI for environments where the stakes are high and the users are often stressed. Getting Connection right is not optional in those environments. It is the whole job.
