Research shows AI trust degrades over time and doesn't recover when systems improve. This isn't an AI problem, it's a UX problem. And it's compounding.
The sticky trust problem is worse in enterprise than individual use. When an AI system gets something wrong in a high-stakes retail moment - a pricing decision or an inventory call - the organisation doesn't narrow its use the way an individual does.
It adds governance layers and manual overrides that become permanent. The system keeps running but trust never recovers because nobody defined what recovery looked like before the failure happened.
The guardrail added in response to one bad output in January is still there in December, long after the model improved.
The sticky trust problem is worse in enterprise than individual use. When an AI system gets something wrong in a high-stakes retail moment - a pricing decision or an inventory call - the organisation doesn't narrow its use the way an individual does.
It adds governance layers and manual overrides that become permanent. The system keeps running but trust never recovers because nobody defined what recovery looked like before the failure happened.
The guardrail added in response to one bad output in January is still there in December, long after the model improved.
This is a great angle! I guess we are all still trying to figure out the best ways to adapt while systems are also adapting
“Trust updates slowly and requires many contradictory experiences to shift”. A powerful reminder for product teams.
Thank you for sharing your thoughts on this subject.