AI in Scientific Experimentation
Search documents
Ginkgo Bioworks' Autonomous Laboratory Driven by OpenAI's GPT-5 Achieves 40% Improvement Over State-of-the-Art Scientific Benchmark
Prnewswire· 2026-02-05 19:00
Core Insights - Ginkgo Bioworks, in collaboration with OpenAI, has developed an AI system that autonomously designs and executes biological experiments, achieving a 40% reduction in cell-free protein synthesis reaction costs compared to the current state of the art [1][4][7] Group 1: AI and Autonomous Lab Integration - The study showcases a practical application of Ginkgo's autonomous lab, utilizing OpenAI's GPT-5 model to design, execute, and analyze experiments in a closed-loop workflow [3][5] - The autonomous lab executed over 36,000 experimental conditions across six iterative cycles, generating nearly 150,000 experimental data points [5][7] - Human involvement was limited to reagent preparation and system oversight, while the AI handled experimental design and data interpretation [5][6] Group 2: Cost Reduction and Efficiency - The autonomous lab produced a benchmark protein, superfolder green fluorescent protein (sfGFP), at a cost of $422 per gram, down from the previously reported $698 per gram, marking a 40% cost reduction [4][7] - The integration of AI in the lab is expected to lead to more experiments being conducted, as lower reagent costs will facilitate increased data generation and scientific progress [4][8] Group 3: Commercial Potential - Ginkgo is now offering the AI-optimized reaction mix for sale in its reagents store, indicating the commercial viability of AI-driven scientific advancements [7][8] - The Pydantic model used for validating experiments will be released as open source, further promoting transparency and collaboration in scientific research [6][8]