This study investigated the impact of artificial intelligence (AI) and peer interactions on the speaking performance and attitudes of EFL university students using a withinsubjects design. Thirty-nine participants engaged in both AI-based and peer-based speaking activities over 14 weeks and completed pre- and post-speaking tests, as well as pre-, mid-, and post-questionnaires. At the end of the semester, they also submitted reflective essays describing their experiences. Results revealed significant improvements in both speaking performance and self-perceived speaking skills from pre- to post-test. However, students’ attitudes toward AI and peer interactions diverged. Engagement, interest, and participation in AI interactions declined over time, whereas these affective dimensions in peer interactions remained stable or improved. Enjoyment and concentration were also consistently higher in peer interactions than in AI interactions. Overall, the findings suggest that although AI can enhance EFL learners’ speaking performance through repetitive practice and timely feedback, peer interactions play a more enduring role in supporting engagement, enjoyment, and meaningful language learning over time.

1. INTRODUCTION
2. LITERATURE REVIEW
    2.1. Positive Impact of Conversational AI on Speaking Skills and AffectiveDimensions
    2.2. Reconsidering the Effectiveness of AI Technologies
    2.3. The Current Study
3. METHODOLOGY
    3.1. Participants
    3.2. Treatments: AI Interactions Versus Peer Interactions
    3.3. Instruments
    3.4. Procedures
    3.5. Data Analysis
4. RESULTS
    4.1. The Impact of AI- and Peer-based Speaking Practice on ActualSpeaking Performance and Self-perceived Speaking Performance
    4.2. Students’ Attitudes Toward AI- and Peer-based Speaking Practice
    4.3. Students’ Perceptions of AI-based Versus Peer-based Speaking
5. DISCUSSION
6. CONCLUSION
REFERENCES

• Juhee Lee(Associate Professor, Department of English Education, Hongik University, 94 Wausanro, Mapo-gu, Seoul 04763, Korea)