AI-powered research boosts efficiency by 3,000 times

The University of Toronto, in collaboration with Harvard, MIT, and 17 other institutions, has utilized AI technology to complete a systematic review study in just two days—something that would traditionally take 12 years. Named otto-SR, this AI workflow combines GPT-4.1 and o3-mini to automate literature screening and data extraction, achieving an efficiency boost of up to 3,000 times. Research shows that otto-SR outperforms traditional human review in both sensitivity and accuracy, identifying more relevant studies and demonstrating significant potential in the field of evidence-based medicine. This breakthrough has garnered widespread attention in the academic community, signaling the important role AI is poised to play in scientific research in the future.