In a video posted on Twitter, machine learning engineer Santiago Valdarrama showed how he and a colleague used the hugely popular OpenAI tool to program the robot dog Spot to verbally answer the system’s questions.

“These robots perform automated tasks every day”Valdarrama explained on Twitter, noting that each task contains “miles of complex configuration files” that only engineers can understand.

This is where ChatGPT comes into play. We show him the configuration files and the task results. Then we ask questions using this context. Combine that with a voice interface and we have a great way to request data!
– says the programmer.

As seen in the video, Spot even adds gestures to his answers: nodding to say “no” or bowing to say “yes”. Valdarrama integrated text-to-speech technology and fallback. As a result, the robot dog hears, understands and analyzes human voice requests. When the neural network generates a text response, it is converted into a sound and the dog responds aloud to the person.

While the interface itself may seem innocuous enough, many who watch the video are terrified of it and claim they are even closer to a robot uprising. In one of the posts on Twitter, the user added a gif of a robot with a machine gun and wrote: “One day we will look back and mark today as a turning point”.

It is worth noting that the integration of artificial intelligence into robots has always been part of the plan. Having signed a multi-billion dollar deal with OpenAI, Microsoft released a document in February outlining the design principles for integrating ChatGPT into robotics.