Abstract
Artificial intelligence (AI) is changing the field of synthetic biology research. This approach has the potential to make the design-build-test-learn (DBTL) cycle more predictable by improving design prioritization, experimental planning, and data analysis. This review aims to analyze the application of artificial intelligence (AI), including machine learning (ML), deep learning (DL), and language models, across synthetic biology, specifically neural networks for genetic circuit design, protein engineering, metabolic pathway optimization, and multi-omics data analysis. The main focus is to identify how AI tools are applied across the DBTL cycle, from design to final data analysis, and to assess their validation and evidence levels. The reviewed studies indicate that, in specific applications, AI-based methods can reveal patterns and relationships that may be difficult to identify using conventional approaches, while deep learning (DL) and language models can reveal patterns and relationships that may be difficult to identify using conventional approaches. These studies also indicate that AI methods can improve prediction performance and reduce redundant experiments in specific applications. Furthermore, AI can support high-throughput technologies and has potential for genome-scale design. However, important limitations remain, including uncertainty in model behavior, limited data validation and transparency, difficulties in data analysis and interpretation, uncertainty in model predictions, and safety and ethical concerns. Therefore, the safe and widespread adoption of AI technologies in the healthcare, energy, and environmental biotechnology sectors requires the development of stronger regulatory and ethical standards.