This study aims to analyze interpreters’ performance in handling numbers, proper nouns, and complex sentences during AI-assisted simultaneous interpreting. By integrating AI speech recognition capabilities, an eye-tracking experiment was conducted with ten master’s students majoring in English interpretation at Shanghai International Studies University. The experiment employed eye-tracking technology to explore the eye movement behaviors and interpreting performance of interpreters when faced with numbers, proper nouns, and long sentences under the assistance of AI real-time subtitles. The results indicate that AI real-time subtitles effectively support interpreters in segments containing numbers and proper nouns; however, the intervention of subtitles did not significantly enhance the interpreters’ performance in handling long and complex sentences. This study also discusses potential optimization directions for AI real-time subtitle technology and its implications for interpreting education.