You can tell that we’re going to end up talking with cats and dogs.
You know how there were tricorders in Star Trek and then, decades later, we got iPhones?
If there were a scale for Cultural readiness level to parallel Nasa’s Technological readiness level, where TRL 1 is Basic principles observed
and TRL 9 is competitive manufacturing
, then CRL 1 would be “in fiction” – Star Trek basically.
And CRL 2 would be “on TikTok” – making its way up the scale.
Anyway so there’s a talking dog on TikTok.
It doesn’t really talk. It uses “Dog Buttons”: big plastic buttons that your dog can push with its paws, and a voice says a word, so the dog can say Kenny want treat
or whatever.
This technique pioneered (I understand) by speech-language pathologist Christine Hunger and her dog Stella. Their story:
Since dogs can understand words, could Stella use an AAC [Augmentative and Alternative Communication] device to express herself the same way my patients did?
(That link has videos.)
The dog can now express 45 words with combinations of up to 5 words. (More over at the Guardian (2020).)
Hunger’s site also has a shop where you can buy (a) their book, and (b) Dog Buttons.
CRL 3 would be where the technology appears on Amazon. Here’s a comparison shopping guide to the best:
- light-up dog buttons
- potty buttons
- dog button training programs
This is dumb, right? It’s just a soundboard being used so your dog can call you from across the house? It’s no different from tricks like roll over or shake hands?
Well. Kinda. Maybe.
BUT: cultural readiness.
Cultural desire creates the conditions for future technology research and development. And now we have AI… well if AIs can do protein folding then why not barks and yelps to English?
TANGENTIALLY:
- That time this guy I know’s cat figured out how to get him to open the front door utilising the Bitcoin network
- That time when jazz legend Charles Mingus taught his cat how to sit on the seat and use the indoor loo, and wrote a book about it
- That time in 1973 where the BBC TV show That’s Life! famously broadcast a dog that could say
sausages
and it turns out that the dog was discovered and filmed by a then-young avant-garde conspiracist film-maker Adam Curtis
- It turns out there is a sequel to 101 Dalmatians and it is wild. The Starlight Barking:
after a strange euphoric moment followed by a moment of terror, Sirius, the Lord of the Dog Star, appears on Nelson’s Column. He explains to all the dogs that he is lonely and is offering them the chance to avoid the pain of possible nuclear war in the future.
Ok.
Cultural readiness is one thing. Technology is another.
AND SO:
There are already organisations researching AI to speak with whales and animals generally:
- Project CETI is
applying advanced machine learning and gentle robotics to listen to and translate the communication of sperm whales.
– the current milestone on the scientific roadmap is to gather training data in the form of whale behaviour and vocalisations
- Earth Species Project is aiming to use
a new machine learning technique that could learn a geometric representation of an entire language
to translate between the latent space of English and the latent space of, say, corvids. Current research includes how to synthesise vocalisations of crows and humpback whales.
The New York Times covers the projects of each in more detail – the goal is ambitious:
“Let’s try to find a Google Translate for animals,” said Diana Reiss, an expert on dolphin cognition and communication at Hunter College and co-founder of Interspecies Internet, a think tank devoted to facilitating cross-species communication.
(Paywall-busting link here. Read the whole thing.)
Project CETI is planning to use a 28 underwater microphones and AV-enabled robot fish to record whales acoustics.
“Every which way we turn there’s another question,” said David Gruber, a marine biologist at Baruch College who leads Project CETI. “If there was a big event that happened a week ago, how would we know that they’re still communicating about it? Do whales do mathematics?”
And, on a species of crow which is at risk of extinction:
“They keep them in these aviaries to breed birds for future releases. But what if these crows no longer know how to speak crow?”
So how long do we give it before this technology is a reality – 20 years? 30 years? The bottleneck does seem to be recording training data, for the moment.
Objection #1 is that focusing on acoustics seems a bit… human-provincial maybe? Like: dogs have great capacity to smell and to, uh, generate smell. Won’t that be as much part of that vocabulary as anything else? Or whales: perhaps cetaceans speak in water vortices as much as clicks and whistles.
More training data required, I suspect.
Maybe in this respect cheap, lo-fi electronic Dog Buttons are better than AI whale-song, in that they create a new trading language instead of interpreting existing sounds?
This was the goal of CHAT by the Wild Dolphin Project or this underwater keyboard at Epcot in Florida, also for human-dolphin communication.
Perhaps we’ll end up having to co-create new languages.
Objection #2 is that, well, we already speak with animals, as anyone who hangs out with animals knows.
There is very little misunderstanding when my cat speaks to me or I speak to my cat, for example.
HOWEVER:
Some people can speak Chinese. I can’t. When Google Translate came along, suddenly we were able to email directly with factories in Shenzhen and speak - through copy-and-pasted green text machine translation - with reps on the floor instead of via agents. It unlocked manufacturing for us, a small design firm in London in the early 2010s.
So I wonder what the parallel is? Strangers!
It would be great to be able to speak with dogs, cats, and crows in their native languages, without having to co-create languages.
Sure, me and my cat live together and so we’ve figured out how to have a conversation – she’s silent much of the time. Do cats gossip? That’s what us humans fill the time with.
But in species with low cultural transmission, where learnt languages can’t be passed on, native tongue translation would mean I could talk with non-human strangers.
Wouldn’t it be great to say hi to a dozing dog on the street and mutually give appreciation to the hot sun?
Or spot a crow on the wall, and ask it if it knows the way to the nearest train station, or a coffee shop, and so on?
You can tell that we’re going to end up talking with cats and dogs.
You know how there were tricorders in Star Trek and then, decades later, we got iPhones?
If there were a scale for Cultural readiness level to parallel Nasa’s Technological readiness level, where TRL 1 is and TRL 9 is , then CRL 1 would be “in fiction” – Star Trek basically.
And CRL 2 would be “on TikTok” – making its way up the scale.
Anyway so there’s a talking dog on TikTok.
It doesn’t really talk. It uses “Dog Buttons”: big plastic buttons that your dog can push with its paws, and a voice says a word, so the dog can say or whatever.
This technique pioneered (I understand) by speech-language pathologist Christine Hunger and her dog Stella. Their story:
(That link has videos.)
The dog can now express 45 words with combinations of up to 5 words. (More over at the Guardian (2020).)
Hunger’s site also has a shop where you can buy (a) their book, and (b) Dog Buttons.
CRL 3 would be where the technology appears on Amazon. Here’s a comparison shopping guide to the best:
This is dumb, right? It’s just a soundboard being used so your dog can call you from across the house? It’s no different from tricks like roll over or shake hands?
Well. Kinda. Maybe.
BUT: cultural readiness.
Cultural desire creates the conditions for future technology research and development. And now we have AI… well if AIs can do protein folding then why not barks and yelps to English?
TANGENTIALLY:
Ok.
Cultural readiness is one thing. Technology is another.
AND SO:
There are already organisations researching AI to speak with whales and animals generally:
The New York Times covers the projects of each in more detail – the goal is ambitious:
(Paywall-busting link here. Read the whole thing.)
Project CETI is planning to use a 28 underwater microphones and AV-enabled robot fish to record whales acoustics.
And, on a species of crow which is at risk of extinction:
So how long do we give it before this technology is a reality – 20 years? 30 years? The bottleneck does seem to be recording training data, for the moment.
Objection #1 is that focusing on acoustics seems a bit… human-provincial maybe? Like: dogs have great capacity to smell and to, uh, generate smell. Won’t that be as much part of that vocabulary as anything else? Or whales: perhaps cetaceans speak in water vortices as much as clicks and whistles.
More training data required, I suspect.
Maybe in this respect cheap, lo-fi electronic Dog Buttons are better than AI whale-song, in that they create a new trading language instead of interpreting existing sounds?
This was the goal of CHAT by the Wild Dolphin Project or this underwater keyboard at Epcot in Florida, also for human-dolphin communication.
Perhaps we’ll end up having to co-create new languages.
Objection #2 is that, well, we already speak with animals, as anyone who hangs out with animals knows.
There is very little misunderstanding when my cat speaks to me or I speak to my cat, for example.
HOWEVER:
Some people can speak Chinese. I can’t. When Google Translate came along, suddenly we were able to email directly with factories in Shenzhen and speak - through copy-and-pasted green text machine translation - with reps on the floor instead of via agents. It unlocked manufacturing for us, a small design firm in London in the early 2010s.
So I wonder what the parallel is? Strangers!
It would be great to be able to speak with dogs, cats, and crows in their native languages, without having to co-create languages.
Sure, me and my cat live together and so we’ve figured out how to have a conversation – she’s silent much of the time. Do cats gossip? That’s what us humans fill the time with.
But in species with low cultural transmission, where learnt languages can’t be passed on, native tongue translation would mean I could talk with non-human strangers.
Wouldn’t it be great to say hi to a dozing dog on the street and mutually give appreciation to the hot sun?
Or spot a crow on the wall, and ask it if it knows the way to the nearest train station, or a coffee shop, and so on?