This investigation analyses the errors that automatic subtitling systems produce when generating subtitles of a Jordanian Arabic dialectal variant that is spoken in the district of Ayy in the governorate of Al-Karak, which is located in the South of Jordan. A new categorization for the subtitles generated by Artificial intelligence (AI) is proposed. The errors are sorted into two main categories, minor errors, and major errors, and are related to deletions, substitutions, or insertions. The quantitative analysis measures the accuracy rate and word error rate (WER) to reveal that the performance of Veed’s ASR system is approved to be better than YouTube’s ASR system, as the accuracy rate for Veed is 54%, and for Youtube is 47%, while WER for Veed is 46% and for Youtube is 53%. More research is needed for the ASR systems and Arabic dialects since such investigations can help improve the ability of these systems to recognize more dialectal Arabic words and overcome linguistic challenges that may face these systems.