This white paper examines how AI-assisted feedback is perceived compared to feedback from subject-specific and out-of-field instructors in design education. Using a quasi-experimental field design at Macromedia University (Hamburg campus), we studied two cohorts in the winter terms 2023/24 and 2024/25. Each group received three standardized, blinded feedback texts (subject-specific professor, out-of-field instructor, AI) in rotated order. Students rated how helpful, appropriate, and motivating they found the feedback texts on 5-point Likert scales. Findings: Human feedback—especially from subject-specific professors—was consistently rated higher on a majority of the dimensions. Surprisingly, students perceived no reliable differences on the helpful dimension. Thus, AI feedback is a useful component, particularly for rapid, structured first-pass assessments, but it does not substitute domain expertise when precise content alignment is required. Recommendations: Use AI to pre-structure and provide initial cues; rely on instructors for precise, contextualized feedback and motivation. Limitations include a small sample and a specific course context. The next step is a new quasi-experimental field design focusing on the how students use and perceive AI tools in their workflow.