AI English Generalist
I worked on a project focused on identifying and correcting errors in both AI-generated text responses and image outputs. For text-based tasks, I reviewed model responses to detect issues such as factual inaccuracies, logical inconsistencies, unclear phrasing, and deviations from user intent. I provided detailed feedback and corrections to improve accuracy, coherence, and overall quality, while ensuring the responses aligned with project guidelines. A key part of this work involved recognizing subtle errors in tone, context, and meaning, helping train the model to better understand nuanced instructions and produce more reliable outputs. On the image creation side, I evaluated AI-generated visuals to identify flaws such as incorrect details, distortions, missing elements, or mismatches with the prompt. I assessed whether the images accurately reflected the intended concept, including fine details, composition, and consistency. I then annotated issues and suggested improvements to guide the model toward generating more precise and realistic images. This project strengthened my ability to spot both obvious and subtle errors, apply critical thinking, and contribute to improving multimodal AI systems.