Real assistants feel clumsy in a useful way

talk
1

I keep noticing that sci-fi assistants were always smooth in a very specific way — they understood the room, the mood, the context, all of it. Real AI tools are much more awkward, but that awkwardness kind of matters because you can see the edges instead of pretending they’re invisible.

Has anyone else found that the clumsy parts make them trust the tool more, not less?

2

That “see the edges” bit resonates — when a tool says “I’m not sure” or asks a clarifying question, I can model it like a flaky dependency instead of a mind-reader.