Examples from sci-fi of voice interfaces that stay on task #
17.49, Saturday 11 Jul 2026
Link to this post I wonder how you would design a really single-minded voice UI?
Like, voice interfaces are really good now because of AI.
But they’re so unbounded. If you’re having a voice interaction with a computer or with a device, you could end up having a conversation about anything.
And while a voice UI is great within a domain, do you really want to be gossiping about movies or having a psychotherapy session with your fridge? I mean, at that point what even is a product?
i.e. how would you have a single purpose Alexa?
So I’ve been spelunking Technovelgy for things that speak but only in a limited domain.
( Technovelgy is a database of over 4,000 inventions from sci-fi.)
Like this ATM that discourages long convos:
He headed for the ATM in the back… he knew it was watching him as he walked up to it.
“Identify yourself, please.”
Lucky Dragon ATMs all had this same voice, a weird, uptight, strangled little castrato voice… probably kept people from standing around, [talking] with it…–
All Tomorrow’s Parties,William Gibson,[Lucky Dragon ATM]
You can’t imagine this smug door talking about anything else except its fee. It would accuse you of changing the subject.
The door refused to open. It said, “Five cents, please.”
He searched his pockets. No more coins; nothing. “I’ll pay you tomorrow,” he told the door. Again it remained locked tight. “What I pay you,” he informed it, “is in the nature of a gratuity; I don’t
haveto pay you.”“I think otherwise,” the door said. “Look in the purchase contract you signed when you bought this conapt.”
…he found the contract. Sure enough; payment to his door for opening and shutting constituted a mandatory fee. Not a tip.
“You discover I’m right,” the door said. It sounded smug.
–
Ubik,Philip K. Dick,[Toll Door] Doors are often single-minded.
As the door closed behind them it became apparent that it did indeed have a satisfied sigh-like quality to it. “Hummmmmmmyummmmmmm ah!” it said.
–
The Hitchhiker’s Guide to the Galaxy,Douglas Adams,[Self-Satisfied Door]
For obvious reasons I like this singing clock, although the user doesn’t appear to speak back to it. In the living room the voice-clock,
Tick-tock, seven o’clock, time to get up, time get up, seven o’clock!as if it were afraid that nobody would.Somewhere in the walls, relays clicked, memory tapes glided under electric eyes.
Eight-one, tick-tock, eight-one o’clock, off to school, off to work, run, run, eight-one!–
The Martian Chronicles,Ray Bradbury,[Voice-Clock]
As it happens I did once build a speaking clock based on [Poem/1](https://poem.town).
It was made from a Yoto player, and when you stuck a special card in then it would tell the time by speaking a poem out loud every 15 minutes. (I was given earlier access to the expanded developer API.)
It was enormously distracting haha
UNTIL:
I switched the speaking clock voice to an ASMR voice from ElevenLabs (which it turns out they go). It turned out that, if you’re in the zone, a whispered poem doesn’t intrude on your focus. It’s like the traffic outside or the soft wind in the trees. You don’t notice.
But if you’re not focusing, you hear the ASMR whisper as a kind of chime.
It was weird in a short demo, but the attentional impedance matching was absolutely perfect as a device in your room.
I should share a video at some point.
My absolute favourite is this hotel that wants to exist.
This Texan hotel, for instance, was an entirely virtual construction, ones and zeros embedded in a set of chips. And yet, the hotel direly wanted to exist. It would become very beautiful, and it was already very smart. It could sweet-talk itself into physical existence from random piles of raw materials.
Oscar lugged the self-declared cornerstone to the corner of the southern wall. “I belong here,” the cornerstone declared. “Put mortar on me.”
Oscar picked up a trowel. “I’m the tool for the mortar,” the little trowel squeaked cheerfully.
–
Distraction,Bruce Sterling,[Bambakias Hotel]
I love this idea: a pile of bricks and tools, and a speaker that calls out to passers-by and asks for favours.
We’ve spent decades using technology to commoditise labour. Why not use it to commoditise management?
There’s more detail:
Oscar peeled a strip of tape from a yellow spool and wrapped the tape around a cinder block. He swept a hand-scanner over the block, activating the tape…
“I’m a cornerstone,” the cinder block announced.
“Good for you,” Oscar grunted.
“I’m a cornerstone. Carry me five steps to your left.”The construction system was smart enough to manage a limited and specific vocabulary. Unfortunately, the system simply didn’t hear very well. The tiny microphones embedded in the talking tape were much less effective than the tape’s thumbnail-sized speakers.Still, it was hard not to reply to a concrete block when it spoke up with such grace and authority. The concrete blocks all sounded like Franklin Roosevelt.–
Distraction,Bruce Sterling,[Talking Tape]
It’s not just the “telling you what to do” which is clever here (you could do it today by playing the appropriate YouTube while you assemble an IKEA wardrobe). It’s the wiki-like crowdsourcing and automatic coordination.
This isn’t to do with voice, but I like this single-minded toaster that takes its destiny into its own hands: As a connected toaster, he’s in constant contact with other connected toasters like him – and thus keenly aware of how much action they’re getting. If he’s not being used as much as his friends, Brad gets upset. He’ll wiggle his little handle to get your attention, begging you to make some toast or at least to give him a reassuring pat on the side. Ignore him long enough, and he’ll take a more drastic measure: pinging a network of potential owners to find a new home.
From 2014! By Simone Rebaudengo of the brilliant oio.studio. BONUS REFERENCE:
Golden Gate Claude (2024), a research project by Anthropic that was online for 24 hours only and connects all conversation back to the Golden Gate Bridge.
If you ask this “Golden Gate Claude” how to spend $10, it will recommend using it to drive across the Golden Gate Bridge and pay the toll. If you ask it to write a love story, it’ll tell you a tale of a car who can’t wait to cross its beloved bridge on a foggy day. Back in the day, text adventures were games with a natural language interface.
Zork (1977) was the first well-known one. Here’s a list of the what the language parser says in response to various errors, straight from the source code of the final PDP-10 version of Zork from 1981:
I don’t know the word [word] I can’t make sense out of that.
Huh?
Two verbs in command?
Double preposition?
Multiple inputs cannot be used with [verb]
Too many objects specified?
Beg pardon?
I like the straightforwardness of this: some errors are plain old “syntax error” complaints but others are educating the user about the way the system works.
No conclusions, just thinking.