| |
 |
| Paper: |
AI Co-Scientist for the Habitable Worlds Observatory: Enhanced Biosignature Detection by Multi-Agent Systems |
| Monograph: |
11, HWO25 Proceedings Part II: Mission Framework, Technology, and Broader Contributions |
| Page: |
373 |
| Authors: |
Avi Shporer and Iddo Drori |
| DOI: |
10.26624/RGDB2784 |
| Abstract: |
We present an artificial intelligence (AI) co-scientist that enhances HWO's scientific capabilities by performing autonomous and reliable research, including hypothesis generation, data analysis, and adaptive mission planning. Using a generate-evolve-review loop, AI agents continuously propose and refine hypotheses by analyzing observational data, reviewing and evolving the system and its research. Our system achieves 96.2% overall accuracy in biosignature detection on synthetic spectral data, with a false positive rate of 2.6% and false negative rate of 4.8%. The framework processes multi-wavelength observations in real-time, identifying weak atmospheric signals that conventional pipelines miss. Using Bayesian inference and deep learning, the AI co-scientist enables dynamic mission adaptation and optimizes observational strategies based on reliable intermediate findings. This human-AI collaboration enhances HWO's potential for discovering extraterrestrial life. |
|
|
|
|
|
