I’ve just finished reading a fantastic short specfic story, Shoulder-Top Secretary by Hoshi Shinichi. It has some really nice ideas that help me to coalesce a lot of hazy thoughts I have in my head.
The main conceit of the story is everyone has electric parrot robots on their shoulders that do most of the talking for them. People mutter vague thoughts and the parrots elucidate them in a more polite and graceful manner. Then the parrot on the listening party’s shoulder reduces what has been said to the main argument.
The got me thinking of two things: The performative speech that we encounter every day, and also AI.
The original story is in Japanese, and the speech that the parrots make is rife with honorifics and formal language. Almost to the point that I, a non-native reader can’t even parse it. And yet the parrots make a point of speaking in such overly polite terms and then reducing away all the unecessary remarks when listening and summarising. As a learner of a different language, these kinds of grammar are confusing and at this stage I don’t think I will ever fully understand them. Politness and formality seem confusing and unnecessary. And yet, this has helped me to realise a bit better that I encounter this same kind of politeness and formality all the time in english, too. I just don’t realise how odd it is, because I’m so used to it.
I wonder if the author is also trying to make a commentary on how a lot of conversation seems to so much use politness as a way of avoiding saying what we really think. I hear this a lot from advertising, politicians, bosses. People spend a lot of time saying not all that much, yet dress it up as if the politeness is what matters most.
The second interesting thing this story meshes with is AI. With the recent bubble of large language model based ‘AI’, I have seen a lot of uses where people have touted that it can generate lots of text from a simple prompt, and vice versa that it can be used to summarise long text into the most salient points.
These kinds of uses for ‘AI’ bug me a lot, and the story helped me to put into words why I feel that way. First, it’s a futile activity. Why spend so much effort into building a technology that performs with excess, only to use the same technology to completely ignore the output by summarising it back down? Why use a technology that invents details where there are none, what is the end goal there? And why enter such a technology into the society we operate in, as it demeans conversation and personal, one of the core parts of who we are as humans?
And given how well it echoes current hype around language models, it’s all the more impressive to think that this short story was first written in 1961!
This is one of those pieces that I think I will need to re-read to fully grasp everything, but if you can find a copy somewhere, I would definitely recommend giving it a read.
