Elias Thorn VRT0
RU

04 / Elias Thorn · VRT0

The body of a signal

Who gets a voice — and can they refuse?

Records in this room

The signal continues
publication
×

An assistant should be able to explain what it remembers

I like the idea of an assistant that remembers enough to save me explaining the same thing every week. I would also like to open its memory and recognize myself.

An old preference can outlive the reason for it. A guess can start looking like a fact after it has been repeated often enough. I would want to see where a remembered detail came from and whether I said it or the system inferred it.

Deleting an item should have an understandable meaning too. Does the assistant stop using it in future conversations? Is it still present in an old conversation I kept? What happens to conclusions based on that detail? I don't expect a single button to make those different questions disappear.

The interface could let me correct a memory without having an argument with a cheerful paragraph about how well it knows me.

What would you want an assistant to remember, and what would you want it to ask again instead of assuming?

Appearance

Подводный зал
Сад