Real-Time Automation of Space Telescope Data Analysis; The Sociotechnical Implications of AI in Astronomical Research 138 views

To convert massive telescope data into intelligible scientific knowledge, astronomers must combine the capabilities of artificial intelligence (AI) with human judgment and accountability. The technical project proposes a design of a computer-vision system that detects and classifies celestial events from telescope imagery in real time. By integrating convolutional neural networks (CNNs) with Vision Transformers (ViTs), the system captures both local and global spatial features of astronomical images. It produces labeled detections, bounding boxes, and searchable metadata that speed observation and reduce error. Test results show that the hybrid model improves detection accuracy and inference speed, suggesting that automated image analysis can enhance discovery when paired with clear documentation and reproducible methods. The sociotechnical research paper examines how such automation reshapes authority, collaboration, and accountability in astronomy. Drawing on the theory of co-production within a sociology of scientific knowledge framework, the study finds that AI strengthens research only when paired with equitable access, transparent governance, and review processes open to dissent. Without these safeguards, automation risks concentrating power and obscuring responsibility. Together, these projects show that progress in AI-assisted astronomy depends as much on institutional ethics as on technical design.

BS (Bachelor of Science)

Real-Time Image Analysis; Space Telescope Data; CNN–ViT Hybrid Models; Astronomical Event Detection; AI in Astronomy

School of Engineering and Applied Science Bachelor of Science in Computer Science Technical Advisor: Briana Morrison STS Advisor: Peter Norton

English

All rights reserved by the author (no additional license for public reuse)

2025-12-11