site banner

Friday Fun Thread for February 14, 2025

Be advised: this thread is not for serious in-depth discussion of weighty topics (we have a link for that), this thread is not for anything Culture War related. This thread is for Fun. You got jokes? Share 'em. You got silly questions? Ask 'em.

2
Jump in the discussion.

No email address required.

I was talking about LLMs with a Japanese friend the other day, and they brought up a 1963 SF short story by Hoshi Shinichi that neatly prefigured an element of the current ChatGPT debate. Prophetic scifi is always fun, so I figured I'll give a sloppy translation of the abridged representative blurb that everyone seems to quote from it.

"The Secretary on the Shoulder" (from the anthology "Bokko-chan")

It is the near future. A door-to-door salesman is visiting a private residence to pitch his company's product. A single parrot is sitting on the salesman's shoulder, and the resident who just emerged from the entranceway likewise has a parrot perched.

The salesman mumbles to the parrot on his shoulder:

(Buy it)

As he does, the parrot starts speaking fluently.

"I am most sorry for bothering you at such a busy time. Today, I would like to humbly request that I may introduce you to the latest product that is the pride of our company."

The resident's parrot translates this.

(Seems like he's here for sales)

Having received this much, the resident whispers to his parrot.

(Ok, what?)

The parrot in turn translates:

"That will be most welcome. At your pleasure, let's hear about it."

The premise, in more detail, is that everyone has a personal robotic parrot that translates to and from convoluted/polite/considerate/PC prose for the benefit of its owner. The full story features the dialogue between salesman and individual (a housewife) in more detail, where in the end parrot translates the housewife's "I don't need it, so scram already" into a long polite excuse about how she would have to consult with her husband before making a big purchase, and if he would kindly inquire again another time. Then the salesman returns to his office and is chewed out by his boss('s parrot) for not closing more sales, and in turn has his "as if it's that easy lol" turned into a suitably deferential statement of contrition. In the end, he drops by a hostess bar to unwind after work, and is of course welcomed by a wall of flattery from the mamasan's parrot. Closing line: "For [salesman], this is the moment he always looks forward to the most."

Especially during the segments where the prompts turned progressively rude, I couldn't help but think of our occasional posters cracking out the AI rewrite in response to being stop-and-searched by the tone police. Much like in the other modern narrative where LLMs will expand one-liners into compliance forms, powerpoints and press releases which are in turn only ever consumed and reduced into digests by other LLMs, the intention presumably was to make the reader wonder whether all the polite bloat (of which Japanese especially has a lot) really serves any purpose.

It did freak me out when one of my colleagues told me he uses ChatGPT extensively for drafting emails. It occurred to me that at least some of the people he's emailing are also drafting their replies using ChatGPT. What you essentially have is two instances of ChatGPT talking to each other via their respective human puppets intermediaries.

Bruh sound effect