AI for rubric-based grading has gained momentum due to its efficiency and potential for objectivity. However, AI grading systems can exhibit evaluation biases that create inequities that can affect students from diverse backgrounds. First, AI systems struggle to provide consistent and fair evaluations, offering generic feedback that fails to address individual student needs. The absence of transparency in AI decision-making algorithms can exacerbate these challenges, highlighting the urgency for a calibration process to rectify the potential inconsistencies. The purpose of this symposium is to discuss AI grading limitations and observe inconsistencies in value judgments to protect the pursuit for fair and inclusive education.