Computer scientist Santiago Valdarrama owns multiple Spot robot dogs made by Boston Dynamics. A machine learning expert, Valdarrama trains them to conduct inspections. To make it faster to extract data from them, he decided to experimentally kit one out with ChatGPT-driven AI and a voicebox:

Enter a caption (optional)

I found it super-creepy that when asked what its name was, the dog replied "OpenAI" rather than "Spot."

I also think imbuing the robot with "emotion"—where it waggles as if to indicate joy—is a terrible idea. We can run into trouble when we start forgetting these things are machines, and our perception of them having "emotions" is already leading us down a dark path: There are already people who have AI chatbot "girlfriends" that they interact with sexually.

On top of that, we already let AI write its own code, and Google CEO Sundar Pichai himself says that the company doesn't "fully understand" how their own AI comes up with its answers. If we start arming free-roaming robots with AI whose workings are opaque to us, how long until they start developing their own ideas, and leave obedience to real dogs?