I do wonder how many people have had similar discussions with LLMs. Your presentation of the video can do with a little bit of work - you often move your face out of frame, perhaps crop in your lens a little less to account for your natural movement when speaking and reacting to your subject.

The voice models that chat gpt use are pretty dang convincing now! I tend to read faster than I can listen, but this is rather interesting all the same - and... you can tell that it is not full of emoji like its normal output, and em dashes! :P

I wonder if interacting with LLMs over voice get different "outputs" than interacting with them via text, using the same prompt.

Food my brain. More ideas. I don't need them. I then, too, wonder how we react when "listening" rather than reading, and whether it is easier, or harder to distinguish that its an AI through writing, or over voice.