A system can be technically impressive and still be a poor fit for the people expected to use it. Evaluation needs to account for the context around the model: the decisions people make, the consequences of mistakes, and the points where judgment cannot be delegated.
That means bringing stakeholders into the process early, testing against real tasks, and making room for feedback that is qualitative as well as quantitative. It also means defining what responsible use looks like before a system is deployed.
The goal is not to slow innovation down. It is to make sure progress is durable, legible, and worthy of trust.