After a multi-year, multi-billion dollar investment deal with OpenAI, which is behind ChatGPT, Microsoft is stepping up on artificial intelligence. As the software giant continues to integrate OpenAI’s artificial intelligence models into its search engine Bing, its browser Edge and office applications, now robots are next.

Robots will be controlled with ChatGPT

In the article Microsoft published as ChatGPT for Robotics (ChatGPT for Robotics), it talks about how to control robots with ChatGPT and what it will gain by doing so. Microsoft aims to help people interact with robots more easily without needing to learn complex programming languages ​​or details about robotic coding.

“We rely heavily on hand-written code to control robots,” Microsoft researchers say. This is where ChatGPT comes into play. As you know, ChatGPT can also be used for purposes such as coding and decoding. Therefore, it is possible to control robots, robotic systems or drones with normal sentences only with a model trained on software languages ​​and programming.

The team tested this approach by implementing it in several demos. One of these demos involved using ChatGPT to control the drone. Before the test was performed, the chatbot was trained with the necessary commands. The result can be said to be magnificent. As you can see in the video right below, whether it’s an industrial or a home robot, all of them can be intuitively controlled with just sentences.

Sentences turn into codes

I think I can better explain how complex this is with this example: The researchers asked the chatbot to “Take a selfie using a reflective surface.” ChatGPT interpreted this request and brought the drone in front of the mirror and created the necessary code to take a selfie.

Of course, there are some obstacles in front of the application. The query may need to be explained to ChatGPT in detail before the required prompt. However, there is a long way to go and the project is open to development. However, Microsoft states that caution should be exercised when using ChatGPT to control a robot. It is stated that these systems have a tendency to produce wrong and erroneous answers and that human control is very critical.