11 ms·

Squawk bots: Can generative AI lead us to understanding animals?

104 points by agnosticmantis 3y ago

eminence32 3y ago
> At the Earth Species Project, that data comes in the form of sounds, motion, and video recorded in either the wild or captivity, sometimes with accompanying annotations from biologists on what the animal was doing at the time and in what context. Has anything similar been done for human vocalizations (aka spoken language)? Text is obviously so much more efficient, and so much more abundant, so even for human speech this sounds like a challenge.
- aaron695 3y ago
  [dead]
daxfohl 3y ago
Would it be easier to do for baby babble?
- monero-xmr 3y ago
  Once again The Simpsons did it https://youtu.be/aYjlgS-s-Rs https://youtu.be/aYjlgS-s-Rs
vba616 3y ago
If we decide that it does, then conveniently there's no way for the animals to dispute it. I had the same reaction to the idea of legal personhood for natural features of the landscape. Am I too jaded?
- Klaster_1 3y ago
  Maybe uplifted species could assist with the translation effort of their base level counterpart communications?
- krisoft 3y ago
  > If we decide that it does, then conveniently there's no way for the animals to dispute it. Not at all. You can construct tests to see if the animal's behaviour is consistent with what the network say they are "saying". You can go and check if what information you receive through it is consistent with the observed reality. You can test that behavioural differences observed between individuals show up in the translation. For example a dog who you know is hurt, or old is probably less chipper than one who just chased a squirrel. Observers of animals often notice that some individuals are more timid while others are more curious/adventurous. One would expect that their vocalisation also reflects of this difference in personalities. It's not like we just slap a network together and we call it done. If the translator says Lassie is saying that Timmy has fallen down the well you might as well check the well, but if Timmy is not there then this might be a translator failure. > Am I too jaded? Yes.
  - vba616 3y ago
    I agree that "translating" what they are trying to say could result in accurate conclusions. But I don't see what that has to do with my comment. Are you explicitly a radical behaviorist, or do you think animals have internal experience? Suppose that instead of a dog, we were talking about a human with dementia in a nursing home. Do you think that not being able to witness your own condition to a doctor would put you at a disadvantage? In the absence of a feedback mechanism, the virtue and the intelligence of people who are in control of a system do not count for much.
    - krisoft 3y ago
      > Are you explicitly a radical behaviorist, or do you think animals have internal experience? You will have to use smaller words with me, or explain your vocabulary, if we are to have a discussion. > But I don't see what that has to do with my comment You said these things in your original comment: "If we decide that it does" (here I presumed "does" mean AI let us understanding animals) "then conveniently there's no way for the animals to dispute it." That is what I'm disagreeing with. There are many ways animals can dispute us understanding them. Imagine a scenario that we have a gadget which is supposedly translating horse body language and vocalisations to human language. You step up to a big black mare and use the gizmo. It tells you that the horse wants scratches behind its ear. You reach out, but oh horror it was a complete mistranslation. The horse was in distress by you approaching and is completely spoked by you reaching for their ear. It reacts, bites your hand, kicks you and drags you through the mud. That is how you realise that the gizmo's understanding of horse communication is less than perfect. That is the feedback mechanism. You said there is no way, and here is one way they can do it. > Suppose that instead of a dog, we were talking about a human with dementia in a nursing home. Do you think that not being able to witness your own condition to a doctor would put you at a disadvantage? Sure. But animals are not dementia patients and being at a doctor/vet is not the only scenario we can test our understanding of them.
mach1ne 3y ago
I would guess that the sounds animals make are highly context-dependent in a multimodal fashion. We will likely be able to detect some correlations, but decoding larger vocabulary seems unlikely.
- collaborative 3y ago
  Dolphins and orcas seem to be capable of language
  - Y_Y 3y ago
    The classical techniques could do with an update here: https://www.theguardian.com/environment/2014/jun/08/the-dolphin-who-loved-me https://www.theguardian.com/environment/2014/jun/08/the-dolp...
- thih9 3y ago
  Same with people. Tone of voice and nonverbal communication in general is highly context dependent and usually carries more meaning than actual words.
- TheSpiceIsLife 3y ago
  Pairie Dog language https://youtu.be/y1kXCh496U0 https://youtu.be/y1kXCh496U0
tomohelix 3y ago
This is potentially huge. LLMs have shown how they can encompass patterns unknown to its creators and effectively use it. If for some reason, we can actually start seeing something resembling higher thought patterns in animals, that can up end society. But still, we already had lots of evidence how some animals like cetacean and primate are potentially sentient. Many people just don't want to admit it and we can't convince them that human isn't anything special. In the end this may just be another piece of proof on a pile that get ignored.
- IIAOPSW 3y ago
  It is wholly consistent to believe that your actual parrots are also statistical parrots. That's not the same thing as "can't be convinced".
  - AlecSchueler 3y ago
    I've actually stopped believing in my own sentience.
    - feanaro 3y ago
      What does that even mean?
      - nwatson 3y ago
        We live in Skinner boxes.
      - lamp987 3y ago
        LLM-induced psychosis. The new health hazard in the town.
      - postsantum 3y ago
        Not OP, but same here. I feel I am a sentient person only when fully concentrated. But when relaxed, it's stochastic autopilot, both in speech and actions I believe I've seen an article touching this topic here, long before GPT. Will link if I find it
thefoolofdaath 3y ago
`Ethical translation and communication with animals` That's definitely some title from the future :)
taneq 3y ago
Quote from the CIO: "Even before the release of popular generative AI tools like Bard, large language models were helping address highly complex research questions." Was this written in a parallel universe where OpenAI never released an LLM?
- lucubratory 3y ago
  Yes, it's called Google Research :P
pixelpoet 3y ago
Eyyy it's my buddy Aza Raskin :) Met him in 2018 on my first visit to the US, very cool guy. I remember being quite skeptical about his project but these days ML seems more than capable of meeting the challenge.
- tap-snap-or-nap 3y ago
  Sometimes I wish I was in the Americas because it is the center of technological development on this planet but I also like free healthcare and laid back life in Australia, which can feel stagnating at times.
ZunarJ5 3y ago
Isn't there evidence of animal dialect and "accents?" I wonder how this will address that.
- krisoft 3y ago
  > I wonder how this will address that. Same as it already handles "speak like a pirate" and "in the style of Shakespeare" and "Python" and "Hungarian". From the perspective of the AI these are all just "dialects and accents". As long as it will have enough data containing the different accents and dialects it will be able to handle it no problem.
  - notahacker 3y ago
    The issue is that LLMs' handling of "speak like a pirate" is evaluated by humans with a preconceived idea of what coherently expressed pirate-style responses should look like (and their outputs significantly improved by human feedback loops). We also have linters, compilers and interpreters to evaluate correctness of code and reject stuff that's borrowed syntax from the wrong language. I can't see any obstacle to an LLM ingesting a database of squawks and emitting squawks which may or may not be in relevant patterns consistent with how a particular flock of birds communicate, but I can't see the birds advising us on whether it's using them idiomatically or not.
circlefavshape 3y ago
> The plot twist is that we may be able to engage in conversation before we understand what we are saying. Wouldn't this be totally pointless? What's the use of a ChatGPT for elephants if it doesn't allow us to communicate with them?
- lucubratory 3y ago
  We could use the interactions to understand more about them and how they communicate. If we can make "ChatGPT for elephants" (i.e. something that could respond to them with a statistically probable response), we can observe how different communications affect their behaviour much more easily because we could ensure one side of the conversation is always 'talking'. Research like this has already led to us being able to identify various language-like articles from different animals, normally alarm calls because it's very easy to see the response to them. There is also a phenomenon in LLM research where a comparatively tiny amount of a language confers unexpectedly high fluency to a model which is otherwise intelligent in its main language and able to transfer that knowledge over. For example, an LLM needs to be trained on a lot of text to become intelligent in the ways modern LLMs are, which means English text because there's so much of it. But if you then train that LLM on a comparatively tiny amount of another language, like Catalan, it doesn't behave like a model that was trained on that tiny sample of Catalan alone, even when it's only speaking in Catalan. Instead it is nearly as intelligent as it is in English, just in a different language. What that means is that we don't necessarily have an impossible task in front of us in order to get enough human-translatable "language" from any given animal. We may only need to get a surprisingly small amount that we understand for an LLM to be able to generalise and come closer to 'fluency' in that language, which is going to expand the amount of data we have access to by a lot. It's an interesting area of research and I hope it gets expanded on
- famouswaffles 3y ago
  There's no reason to think we couldn't use a chat-GPT for [insert animal] to communicate with them. The idea is that cGPT would be translating for us and vice versa. But that doesn't mean we understanding what is being conveyed exactly.
stiray 3y ago
I doubt we would want to understand the animals based on how we treat them...
osobo 3y ago
I feel this might need some new form of 3D input to get it right. Our grasp of speech made us lazy: The core of what we want to communicate can be captured in sound and therefore words (and hence you can read and understand this). 'Communicating' with my dog is a multi-cue affair, involving pointed looks, head tilts, specific wags and a lot of emoting. I doubt you could consistently pick all that up from (2D) video alone.
- blooalien 3y ago
  I remember reading a while back a study (or theory, or SomeJunkLikeThat™) where dolphins (and possibly whales?) may very well communicate with other dolphins directly in 3D via their sonar capabilities, so that right there could be yet another need for 3D input (and output) in a whole different way than might be expected, presuming they hope to use this idea with dolphins and other sonar capable sea mammals. And then you have bees that communicate with each other through complicated "dances". Fun stuff…
astrange 3y ago
Seems like a pretty big problem that you don't know when any of your data is valid. What if every single whale or cat speaks a different language? It's not like animals have English class.
- api 3y ago
  If they actually have language there are going to be commonalities or it would be worthless as such. There could of course be regional languages or dialects in more intelligent social animals just like humans.
  - astrange 3y ago
    They don't necessarily have shared "language". Cat meows are for one person (their owner) while they use body language and other sounds to communicate with each other.
- inetknght 3y ago
  > What if every single whale or cat speaks a different language? It's not like animals have English class. No, but they have parents. I am sure they have taught languages. Just ask any cat lady who will swear their cats understand them. I'd be more concerned that different regions have different languages and we'd need to control for that.
  - melagonster 3y ago
    this can be more terrible. If we don't let human baby contact to another human, they will lose ability of language. how many puppy take from their mother at first day after birth?
- zenkat 3y ago
  The post goes into this in some detail. In human languages, the embeddings for specific concepts seem to follow very similar geometric relationships no matter what language. If that holds across species, then you could use geometric similarity to map animal embeddings to human concepts. Also, since chatbots just output "meaningful" strings of tokens based on a string of input tokens, you can build a chatbot that can "talk" to animals without us necessarily being able to understand it. But sure how useful that would be, but interesting nonetheless.
hamburglar1 3y ago
ML researchers would get a lot of interesting ideas from Ed Yong's book: "An Immense World: How Animal Senses Reveal the Hidden Realms Around Us" It goes through the extraordinarily diverse sensory systems of all animals. In the push for at-the-edge, exteme-low-latency, low-resource "intelligent" agents, mimicking some of these system may lead to cool systems.
petemir 3y ago
As usual with these type of threads, for those interested in the subject of animal intelligence, I suggest reading "Are We Smart Enough to Know How Smart Animals Are?" by Frans de Waal.
- dandare 3y ago
  Thanks!
jl6 3y ago
Imagine discovering that obligate carnivores and their prey are both sapient. Trolley problems all the way down...
ryanblakeley 3y ago
We should approach animal intelligence and communication with humility. A lot of it is based on smell, body language, and signals we don't pay attention to. I don't think any animal has been shown to tell stories, but obviously they share information in other ways. Only a handful of individuals have demonstrated some kind of ability to understand a story (macaws and koko the gorilla). For anybody interested in animal intelligence, that's the focus of my newsletter.
- jjtheblunt 3y ago
  don't they speak of whales conveying long songs as if stories, largely complex and similar from "telling" to "telling"? what's going on there?
  - ryanblakeley 3y ago
    That's a great place to look for story-telling, I just don't think it's been shown yet (I would love to be wrong about that). If you could decode in their language something about characters taking actions, time and place, past and future, that would be pretty incredible.
- yboris 3y ago
  Related: Are We Smart Enough to Know How Smart Animals Are? by Frans De Waal https://www.amazon.com/Are-Smart-Enough-Know-Animals/dp/0393353664 https://www.amazon.com/Are-Smart-Enough-Know-Animals/dp/0393...
- gerad 3y ago
  Go to another country where you don’t speak the language and try to communicate. You’ll do it with gestures, body language, facial expressions, grunts, tones - not words. I’m guessing that’s a lot closer to how animals communicate than spoken language.
- ddingus 3y ago
  Animals do tell stories. Let's take a dog for example: The dog is out in the wild, away from it's pack. Ever see your dog smell something, or find something it finds interesting? Often, they will rub on it, or roll around in it. They do that so they have a record they can take with them. When said dog returns to their pack, their body holds these records. Their movements communicate their mood and the entire thing is a tale of the day. Another example I had happen was my cat. I would come home and the cat would want to get in the car. Then, if I got up to move, she would put a paw out, like "hold on a moment" and then she would inspect me everywhere, sniffing, looking at me, etc... The cat wanted my story! Where I went, was it the same strange animals and people, or different, when I last ate, and all sorts of stuff. In the evenings, that same cat would come to me, want to cuddle up in the evening and would be near my face, so I tried the same thing and guess what? It's a damn solid story! I now can smell my cat, watch it's movements and get a good sense of it's activity for the day. Back to the dogs. Having returned to the pack, the dog could tell the story of animals it encountered, possible food sources, enemies and other things by the scents it carried home to share with the others. In my view, our biggest struggle understanding animals comes down to both how much they are like us, just in simpler form, and how we fail to see that due to our higher complexity taking us away from basics that make more sense than we know. And with some animals, they are just very different from us, whales, dolphins, octopus and the like. All that said, I agree with you! Humility and an open mind are key to understanding animals better. Frankly, the same tools help us understand one another better.
azubinski 3y ago
Of course not, because in any case, a training set is needed, and it will not be built by the animals themselves, but by people who ask questions like: "Can generative AI lead us to understanding animals?".
kissiel 3y ago
EDIT: wrong thread, sorry.
- titaniumtown 3y ago
  wrong post
grantcas 3y ago
[dead]
machina_ex_deus 3y ago
Just please don't try to understand squirrels, we don't want a dolittle situation...
askin4it 3y ago
They are other nations... " We need another and a wiser and perhaps a more mystical concept of animals. Remote from universal nature and living by complicated artifice, man in civilization surveys the creature through the glass of his knowledge and sees thereby a feather magnified and the whole image in distortion. We patronize them for their incompleteness, for their tragic fate for having taken form so far below ourselves. And therein do we err. For the animal shall not be measured by man. In a world older and more complete than ours, they move finished and complete, gifted with the extension of the senses we have lost or never attained, living by voices we shall never hear. They are not brethren, they are not underlings: they are other nations, caught with ourselves in the net of life and time, fellow prisoners of the splendour and travail of the earth. " Henry Beston, The Outermost House: A Year of Life On The Great Beach of Cape Cod

> At the Earth Species Project, that data comes in the form of sounds, motion, and video recorded in either the wild or captivity, sometimes with accompanying annotations from biologists on what the animal was doing at the time and in what context. Has anything similar been done for human vocalizations (aka spoken language)? Text is obviously so much more efficient, and so much more abundant, so even for human speech this sounds like a challenge.

aaron695 3y ago
[dead]

daxfohl 3y ago

Would it be easier to do for baby babble?

monero-xmr 3y ago
Once again The Simpsons did it https://youtu.be/aYjlgS-s-Rs https://youtu.be/aYjlgS-s-Rs

vba616 3y ago

If we decide that it does, then conveniently there's no way for the animals to dispute it. I had the same reaction to the idea of legal personhood for natural features of the landscape. Am I too jaded?

Klaster_1 3y ago
Maybe uplifted species could assist with the translation effort of their base level counterpart communications?
krisoft 3y ago
> If we decide that it does, then conveniently there's no way for the animals to dispute it. Not at all. You can construct tests to see if the animal's behaviour is consistent with what the network say they are "saying". You can go and check if what information you receive through it is consistent with the observed reality. You can test that behavioural differences observed between individuals show up in the translation. For example a dog who you know is hurt, or old is probably less chipper than one who just chased a squirrel. Observers of animals often notice that some individuals are more timid while others are more curious/adventurous. One would expect that their vocalisation also reflects of this difference in personalities. It's not like we just slap a network together and we call it done. If the translator says Lassie is saying that Timmy has fallen down the well you might as well check the well, but if Timmy is not there then this might be a translator failure. > Am I too jaded? Yes.
- vba616 3y ago
  I agree that "translating" what they are trying to say could result in accurate conclusions. But I don't see what that has to do with my comment. Are you explicitly a radical behaviorist, or do you think animals have internal experience? Suppose that instead of a dog, we were talking about a human with dementia in a nursing home. Do you think that not being able to witness your own condition to a doctor would put you at a disadvantage? In the absence of a feedback mechanism, the virtue and the intelligence of people who are in control of a system do not count for much.
  - krisoft 3y ago
    > Are you explicitly a radical behaviorist, or do you think animals have internal experience? You will have to use smaller words with me, or explain your vocabulary, if we are to have a discussion. > But I don't see what that has to do with my comment You said these things in your original comment: "If we decide that it does" (here I presumed "does" mean AI let us understanding animals) "then conveniently there's no way for the animals to dispute it." That is what I'm disagreeing with. There are many ways animals can dispute us understanding them. Imagine a scenario that we have a gadget which is supposedly translating horse body language and vocalisations to human language. You step up to a big black mare and use the gizmo. It tells you that the horse wants scratches behind its ear. You reach out, but oh horror it was a complete mistranslation. The horse was in distress by you approaching and is completely spoked by you reaching for their ear. It reacts, bites your hand, kicks you and drags you through the mud. That is how you realise that the gizmo's understanding of horse communication is less than perfect. That is the feedback mechanism. You said there is no way, and here is one way they can do it. > Suppose that instead of a dog, we were talking about a human with dementia in a nursing home. Do you think that not being able to witness your own condition to a doctor would put you at a disadvantage? Sure. But animals are not dementia patients and being at a doctor/vet is not the only scenario we can test our understanding of them.

mach1ne 3y ago

I would guess that the sounds animals make are highly context-dependent in a multimodal fashion. We will likely be able to detect some correlations, but decoding larger vocabulary seems unlikely.

collaborative 3y ago
Dolphins and orcas seem to be capable of language
- Y_Y 3y ago
  The classical techniques could do with an update here: https://www.theguardian.com/environment/2014/jun/08/the-dolphin-who-loved-me https://www.theguardian.com/environment/2014/jun/08/the-dolp...
thih9 3y ago
Same with people. Tone of voice and nonverbal communication in general is highly context dependent and usually carries more meaning than actual words.
TheSpiceIsLife 3y ago
Pairie Dog language https://youtu.be/y1kXCh496U0 https://youtu.be/y1kXCh496U0

tomohelix 3y ago

This is potentially huge. LLMs have shown how they can encompass patterns unknown to its creators and effectively use it. If for some reason, we can actually start seeing something resembling higher thought patterns in animals, that can up end society. But still, we already had lots of evidence how some animals like cetacean and primate are potentially sentient. Many people just don't want to admit it and we can't convince them that human isn't anything special. In the end this may just be another piece of proof on a pile that get ignored.

IIAOPSW 3y ago
It is wholly consistent to believe that your actual parrots are also statistical parrots. That's not the same thing as "can't be convinced".
- AlecSchueler 3y ago
  I've actually stopped believing in my own sentience.
  - feanaro 3y ago
    What does that even mean?
    - nwatson 3y ago
      We live in Skinner boxes.
    - lamp987 3y ago
      LLM-induced psychosis. The new health hazard in the town.
    - postsantum 3y ago
      Not OP, but same here. I feel I am a sentient person only when fully concentrated. But when relaxed, it's stochastic autopilot, both in speech and actions I believe I've seen an article touching this topic here, long before GPT. Will link if I find it

thefoolofdaath 3y ago

`Ethical translation and communication with animals` That's definitely some title from the future :)

taneq 3y ago

Quote from the CIO: "Even before the release of popular generative AI tools like Bard, large language models were helping address highly complex research questions." Was this written in a parallel universe where OpenAI never released an LLM?

lucubratory 3y ago
Yes, it's called Google Research :P

pixelpoet 3y ago

Eyyy it's my buddy Aza Raskin :) Met him in 2018 on my first visit to the US, very cool guy. I remember being quite skeptical about his project but these days ML seems more than capable of meeting the challenge.

tap-snap-or-nap 3y ago
Sometimes I wish I was in the Americas because it is the center of technological development on this planet but I also like free healthcare and laid back life in Australia, which can feel stagnating at times.

ZunarJ5 3y ago

Isn't there evidence of animal dialect and "accents?" I wonder how this will address that.

krisoft 3y ago
> I wonder how this will address that. Same as it already handles "speak like a pirate" and "in the style of Shakespeare" and "Python" and "Hungarian". From the perspective of the AI these are all just "dialects and accents". As long as it will have enough data containing the different accents and dialects it will be able to handle it no problem.
- notahacker 3y ago
  The issue is that LLMs' handling of "speak like a pirate" is evaluated by humans with a preconceived idea of what coherently expressed pirate-style responses should look like (and their outputs significantly improved by human feedback loops). We also have linters, compilers and interpreters to evaluate correctness of code and reject stuff that's borrowed syntax from the wrong language. I can't see any obstacle to an LLM ingesting a database of squawks and emitting squawks which may or may not be in relevant patterns consistent with how a particular flock of birds communicate, but I can't see the birds advising us on whether it's using them idiomatically or not.

circlefavshape 3y ago

> The plot twist is that we may be able to engage in conversation before we understand what we are saying. Wouldn't this be totally pointless? What's the use of a ChatGPT for elephants if it doesn't allow us to communicate with them?

lucubratory 3y ago
We could use the interactions to understand more about them and how they communicate. If we can make "ChatGPT for elephants" (i.e. something that could respond to them with a statistically probable response), we can observe how different communications affect their behaviour much more easily because we could ensure one side of the conversation is always 'talking'. Research like this has already led to us being able to identify various language-like articles from different animals, normally alarm calls because it's very easy to see the response to them. There is also a phenomenon in LLM research where a comparatively tiny amount of a language confers unexpectedly high fluency to a model which is otherwise intelligent in its main language and able to transfer that knowledge over. For example, an LLM needs to be trained on a lot of text to become intelligent in the ways modern LLMs are, which means English text because there's so much of it. But if you then train that LLM on a comparatively tiny amount of another language, like Catalan, it doesn't behave like a model that was trained on that tiny sample of Catalan alone, even when it's only speaking in Catalan. Instead it is nearly as intelligent as it is in English, just in a different language. What that means is that we don't necessarily have an impossible task in front of us in order to get enough human-translatable "language" from any given animal. We may only need to get a surprisingly small amount that we understand for an LLM to be able to generalise and come closer to 'fluency' in that language, which is going to expand the amount of data we have access to by a lot. It's an interesting area of research and I hope it gets expanded on
famouswaffles 3y ago
There's no reason to think we couldn't use a chat-GPT for [insert animal] to communicate with them. The idea is that cGPT would be translating for us and vice versa. But that doesn't mean we understanding what is being conveyed exactly.

stiray 3y ago

I doubt we would want to understand the animals based on how we treat them...

osobo 3y ago

I feel this might need some new form of 3D input to get it right. Our grasp of speech made us lazy: The core of what we want to communicate can be captured in sound and therefore words (and hence you can read and understand this). 'Communicating' with my dog is a multi-cue affair, involving pointed looks, head tilts, specific wags and a lot of emoting. I doubt you could consistently pick all that up from (2D) video alone.

blooalien 3y ago
I remember reading a while back a study (or theory, or SomeJunkLikeThat™) where dolphins (and possibly whales?) may very well communicate with other dolphins directly in 3D via their sonar capabilities, so that right there could be yet another need for 3D input (and output) in a whole different way than might be expected, presuming they hope to use this idea with dolphins and other sonar capable sea mammals. And then you have bees that communicate with each other through complicated "dances". Fun stuff…

astrange 3y ago

Seems like a pretty big problem that you don't know when any of your data is valid. What if every single whale or cat speaks a different language? It's not like animals have English class.

api 3y ago
If they actually have language there are going to be commonalities or it would be worthless as such. There could of course be regional languages or dialects in more intelligent social animals just like humans.
- astrange 3y ago
  They don't necessarily have shared "language". Cat meows are for one person (their owner) while they use body language and other sounds to communicate with each other.
inetknght 3y ago
> What if every single whale or cat speaks a different language? It's not like animals have English class. No, but they have parents. I am sure they have taught languages. Just ask any cat lady who will swear their cats understand them. I'd be more concerned that different regions have different languages and we'd need to control for that.
- melagonster 3y ago
  this can be more terrible. If we don't let human baby contact to another human, they will lose ability of language. how many puppy take from their mother at first day after birth?
zenkat 3y ago
The post goes into this in some detail. In human languages, the embeddings for specific concepts seem to follow very similar geometric relationships no matter what language. If that holds across species, then you could use geometric similarity to map animal embeddings to human concepts. Also, since chatbots just output "meaningful" strings of tokens based on a string of input tokens, you can build a chatbot that can "talk" to animals without us necessarily being able to understand it. But sure how useful that would be, but interesting nonetheless.

hamburglar1 3y ago

ML researchers would get a lot of interesting ideas from Ed Yong's book: "An Immense World: How Animal Senses Reveal the Hidden Realms Around Us" It goes through the extraordinarily diverse sensory systems of all animals. In the push for at-the-edge, exteme-low-latency, low-resource "intelligent" agents, mimicking some of these system may lead to cool systems.

petemir 3y ago

As usual with these type of threads, for those interested in the subject of animal intelligence, I suggest reading "Are We Smart Enough to Know How Smart Animals Are?" by Frans de Waal.

dandare 3y ago
Thanks!

jl6 3y ago

Imagine discovering that obligate carnivores and their prey are both sapient. Trolley problems all the way down...

ryanblakeley 3y ago

We should approach animal intelligence and communication with humility. A lot of it is based on smell, body language, and signals we don't pay attention to. I don't think any animal has been shown to tell stories, but obviously they share information in other ways. Only a handful of individuals have demonstrated some kind of ability to understand a story (macaws and koko the gorilla). For anybody interested in animal intelligence, that's the focus of my newsletter.

jjtheblunt 3y ago
don't they speak of whales conveying long songs as if stories, largely complex and similar from "telling" to "telling"? what's going on there?
- ryanblakeley 3y ago
  That's a great place to look for story-telling, I just don't think it's been shown yet (I would love to be wrong about that). If you could decode in their language something about characters taking actions, time and place, past and future, that would be pretty incredible.
yboris 3y ago
Related: Are We Smart Enough to Know How Smart Animals Are? by Frans De Waal https://www.amazon.com/Are-Smart-Enough-Know-Animals/dp/0393353664 https://www.amazon.com/Are-Smart-Enough-Know-Animals/dp/0393...
gerad 3y ago
Go to another country where you don’t speak the language and try to communicate. You’ll do it with gestures, body language, facial expressions, grunts, tones - not words. I’m guessing that’s a lot closer to how animals communicate than spoken language.
ddingus 3y ago
Animals do tell stories. Let's take a dog for example: The dog is out in the wild, away from it's pack. Ever see your dog smell something, or find something it finds interesting? Often, they will rub on it, or roll around in it. They do that so they have a record they can take with them. When said dog returns to their pack, their body holds these records. Their movements communicate their mood and the entire thing is a tale of the day. Another example I had happen was my cat. I would come home and the cat would want to get in the car. Then, if I got up to move, she would put a paw out, like "hold on a moment" and then she would inspect me everywhere, sniffing, looking at me, etc... The cat wanted my story! Where I went, was it the same strange animals and people, or different, when I last ate, and all sorts of stuff. In the evenings, that same cat would come to me, want to cuddle up in the evening and would be near my face, so I tried the same thing and guess what? It's a damn solid story! I now can smell my cat, watch it's movements and get a good sense of it's activity for the day. Back to the dogs. Having returned to the pack, the dog could tell the story of animals it encountered, possible food sources, enemies and other things by the scents it carried home to share with the others. In my view, our biggest struggle understanding animals comes down to both how much they are like us, just in simpler form, and how we fail to see that due to our higher complexity taking us away from basics that make more sense than we know. And with some animals, they are just very different from us, whales, dolphins, octopus and the like. All that said, I agree with you! Humility and an open mind are key to understanding animals better. Frankly, the same tools help us understand one another better.

azubinski 3y ago

Of course not, because in any case, a training set is needed, and it will not be built by the animals themselves, but by people who ask questions like: "Can generative AI lead us to understanding animals?".

kissiel 3y ago

EDIT: wrong thread, sorry.

titaniumtown 3y ago
wrong post

grantcas 3y ago

[dead]

machina_ex_deus 3y ago

Just please don't try to understand squirrels, we don't want a dolittle situation...

askin4it 3y ago

They are other nations... " We need another and a wiser and perhaps a more mystical concept of animals. Remote from universal nature and living by complicated artifice, man in civilization surveys the creature through the glass of his knowledge and sees thereby a feather magnified and the whole image in distortion. We patronize them for their incompleteness, for their tragic fate for having taken form so far below ourselves. And therein do we err. For the animal shall not be measured by man. In a world older and more complete than ours, they move finished and complete, gifted with the extension of the senses we have lost or never attained, living by voices we shall never hear. They are not brethren, they are not underlings: they are other nations, caught with ourselves in the net of life and time, fellow prisoners of the splendour and travail of the earth. " Henry Beston, The Outermost House: A Year of Life On The Great Beach of Cape Cod