What would we hope for from a personality indicator? We'd hope that it would be useful for making predictions.
At an absolute minimum, MBTI could be used to predict the answers people give on an MBTI questionnaire. And what about beyond that? On the basis of someone testing as an I (introvert) vs E (extrovert), we might be able to predict whether someone would feel drained or invigorated after a busy social event. Has there really been work done to show that MBTIs have zero useful predictive value?
I will readily believe that there might be more useful models of personality, and that MBTI is no longer in vogue in psychological literature, and that MBTI was originally motivated on pseudoscientific principles, and that there more principled alternatives — but it doesn't necessarily follow that MBTI cannot be used to make useful predictions.