A new study from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) demonstrates that large language models (LLMs) can be used to generate functional robotic control code directly from natural language instructions. The system, named Code as Policies (CaP), translates high-level tasks described in plain English into executable code for robots with multiple arms and complex toolkits. This approach bypasses traditional, labor-intensive programming methods, allowing for more dynamic and adaptable robotic behavior. The researchers showed that CaP could successfully execute tasks like arranging colored blocks in a specific pattern or coordinating multiple tools in sequence, based solely on written commands. The work highlights a shift toward using the reasoning and code-generation capabilities of LLMs to make robots more accessible and responsive to human operators. For the full details, read the article at https://technologyreview.com/2023/07/12/1076296/robots-code-ai-language-models-mit.