Seeing the Goal, Missing the Truth: Human Accountability for AI Bias