Feedback is a fundamental component in language acquisition, as it is essential for achieving the optimization process required in formal learning contexts. Without feedback, error correction becomes significantly more challenging, often chaotic, and based on incorrect assumptions that, in the worst cases, can lead to the cognitive stabilization of incorrect nodes.

Artificial intelligence (AI), particularly generative AI, introduces new possibilities for automating certain aspects of feedback, enhancing both its precision and accessibility. These tools can more effectively identify formal or structural errors and provide clear explanations along with valid alternatives.

However, for feedback to be minimally effective and productive, it must meet two key objectives: personalization and accurate analysis of error genesis. Personalization refers to the feedback’s ability to adapt to each learner’s specific needs, while error analysis focuses on identifying the causes of errors not only from an interlinguistic perspective (influence of the native language) but also from an intralinguistic perspective (internal processes of the target language).

This study employs a mixed-methods approach to analyze the quality of feedback provided by generative AI tools, focusing on three aspects: detecting formal errors, the clarity of the explanations provided, and the ability to adapt to the learner’s intralinguistic profile. Additionally, the study evaluates the qualitative impact of this feedback from the students’ perspective, focusing on participants enrolled in fully online and asynchronous Spanish as a Foreign Language courses in Universidade Aberta (Portugal).

Finally, the ethical dimensions of using generative AI for student feedback in the teaching and learning process are examined.

Ultimately, this research offers a critical perspective on the role of generative AI in language education, exploring both its benefits and limitations in current educational contexts.

Acknowledgement:
This work is part of the R&D&I project Artificial Intelligence in Language Learning and Teacher Education (CIGE/2023/166).

Keywords: Generative AI, Feedback, Error Analysis, Language Acquisition, Online Education.