<![CDATA[Background: Teacher education programs in the Global South face persistent challenges in implementing fair and context-responsive assessment practices. In southern Colombia, public universities operate within structural constraints including limited funding, heavy workloads, and sociocultural diversity that complicate the adoption of innovative and equity-oriented evaluation models. At the same time, Artificial Intelligence (AI) is reshaping pedagogical decision-making, raising both possibilities and ethical concerns regarding teacher agency, cultural relevance, and data governance.Aims: This study examines how AI-supported alternative assessment practices can enhance reflective learning, autonomy, and critical awareness among pre-service English teachers in southern Colombia while remaining aligned with decolonial and context-sensitive educational principles.Methods: Drawing on a qualitative descriptive design, the research analyzes classroom observations, student-produced assessment artifacts, and reflective journals collected during a 16-week university course on formative and process-based evaluation.Result: Findings indicate that AI-supported assessment expanded students’ multimodal design repertoires, strengthened epistemic agency in their interaction with technological tools, and fostered deeper reflection on validity, coherence, and contextual relevance. However, these developments were mediated by structural constraints, including digital access gaps and practicum realities in under-resourced schools.Conclusion: The study concludes that AI-assisted alternative assessment can foster more equitable and context-responsive learning experiences when integrated dialogically and critically, grounded in decolonial commitments, and attuned to institutional and material realities in the Global South.]]>
