Microsoft and OpenAI have teamed up to explore how the ChatGPT language model can be used to control robots using natural language. The idea is to make it easier for people to interact with robots without having to learn complex programming languages.
The company has published a paper detailing a new set of design principles that ChatGPT uses to give instructions to robots.
The process begins by defining a list of high-level tasks that the robot can perform, writing an instruction that ChatGPT translates into the robots’ language, and then running a simulation in which the robot follows the instructions. The result is adjusted until the robot learns to do it correctly and the code is deployed to the robot.
The risks
While the idea of ​​communicating directly with robots and giving them commands in natural language is appealing, the article points to concerns about the potential risks associated with AI-powered robots. The example of the previous Microsoft experiment that resulted in an unbalanced chatbot propagating racism and plans for world domination is mentioned.
Despite the risks, this collaboration is exciting and represents a breakthrough in robotics. The idea that robots can understand the laws of physics and communicate in natural language is an important goal.
No doubt this technology could make robots more useful in a variety of tasks in different environments.
However, before ChatGPT controls robots in the physical world, it is important to work on security and to resolve any issues that may arise. Microsoft engineers are aware of this and are working to ensure that the technology is secure and reliable. Despite everything, the collaboration between Microsoft and OpenAI is yet another example of the potential that artificial intelligence has to improve and simplify our lives in the future.
